Category: 3. Business

Our Toronto office represented KSV Restructuring Inc., as court-appointed monitor, in the Companies’ Creditors Arrangement Act proceedings of Norwood Sawmills, a manufacturer of outdoor sawmill equipment. At the time of filing, Norwood had approximately $30 million in secured obligations owing to Monroe Capital.

Given the company’s significant liquidity constraints and cross-border operations, a pre-filing sale process was conducted. The transaction was approved and completed within the first few weeks of the proceedings.

This matter highlights our experience advising on complex, expedited restructurings and cross-border transactions.

The team was led by Jennifer Stam.

FRANKFURT, Nov 27 (Reuters) – German drone maker Quantum Systems has raised an additional 180 million euros ($208.67 million) in a new funding round, tripling its valuation to more than 3 billion euros, the company said on Thursday.

Sign up here.

“Triple unicorn status is a testament to our team’s ability to build systems and a company that performs in the most demanding real-world conditions,” said Florian Seibel, the company’s co-founder and co-CEO, in a statement, using a term for private startups worth more than $3 billion.

($1 = 0.8626 euros)

Our Standards: The Thomson Reuters Trust Principles., opens new tab

Introduction

A rapidly evolving ecosystem is developing in health care to provide both patients and professionals with digital solutions for disease management and rehabilitation []. Although evidence of the value of telerehabilitation is increasing for certain neurological disorders such as stroke, to date, there has been little consideration of facial palsy [-]. Bell palsy is the most common single nerve disorder worldwide, leaving patients unable to move muscles on the affected side of their face [,]. It is acknowledged that facial palsy is not a single entity, but rather a feature of different neurological conditions, with Bell palsy representing the majority of cases (approximately 60%) []. Facial palsy will affect 11‐40 people per 100,000 in the population each year, most commonly in the age group of 30‐45 years []. There are approximately 22,500 new facial palsy cases reported each year in the United Kingdom []. Occurrence can be linked to obesity, hypertension, diabetes, upper respiratory conditions, people who are immunocompromised, and pregnancy [,]. Although the underlying cause remains unclear [], facial palsy has long been associated with reactivation of latent herpes virus infections [], with rises reported to be linked to increasing rates of this infection []. More recently, systematic reviews have reported evidence of links with COVID-19 infections and vaccine programs [,]. Over the course of their lifetime, 1 in 60 people will be affected, and 30% will experience incomplete recovery [,]. People with incomplete recovery experience long-term reductions in quality of life, including psychological distress, depression, and social alienation [,-].

Evidence of cost-effective treatments for facial palsy is limited to medication (ie, prednisolone within 72 hours of symptom onset) []. Among nondrug treatments, physical therapies such as facial neuromuscular retraining (fNMR) to optimize muscle resting tone and retrain balanced facial function have been most widely evaluated [-]. Systematic reviews provide evidence of their effectiveness early in recovery and potentially in chronic cases [,,]. However, evidence on cost-effectiveness is lacking. Access to such specialist therapy is also difficult, with 1 in 10 patients in the United Kingdom reporting they travel ≥115 miles for their regular outpatient appointments [].

Early research has suggested that biofeedback directly to patients with facial palsy could improve outcomes and reported interest among patients in innovative digital therapy or telecare [,]. A recent scoping review of telerehabilitation for peripheral facial palsy has identified 18 studies; these include 4 randomized controlled trial protocols, but no completed evaluations []. A similar scoping review of extended reality, including virtual reality (VR), has identified 7 studies but concluded that further research and validation are required []. Facial palsy guidelines do not currently mention telerehabilitation []. In contrast, the evidence base for telerehabilitation in other neurological conditions such as stroke is more well established. A systematic review of stroke home-based rehabilitation has identified over 90 existing systems, including robotics, VR, and game devices, with high effect sizes reported [,]. Robotic gait training is reported to produce a 52% increase in functional ambulation [], video-guided home exercise programs have demonstrated up to 70% mobility gain [,], and VR use in home-based rehabilitation has shown 85% motor recovery gain [,]. While stroke rehabilitation guidelines currently incorporate evidence on the effectiveness of telerehabilitation [], there still remains a significant gap in terms of any evidence on cost-effectiveness [].

This economic analysis of telerehabilitation introduced into the fNMR pathways, therefore, addresses an important evidence gap for neurological conditions. This research is part of a wider program exploring the potential for facial remote monitoring eyewear (Frame) to transform facial palsy therapy []. Frame smart glasses can provide discreet feedback to patients while they undertake prescribed fNMR exercises at home and could improve monitoring by specialist therapists [].

Methods

Summary

The economic analysis included three elements. First, a national budget analysis was used to assess the economic burden associated with facial palsy cases, including treatment costs and the effects of living with acquired facial palsy and long-term facial disfigurement []. Second, a national Delphi exercise was conducted to identify the monetary value placed on various levels of clinical recovery. Third, economic modeling was performed to assess the cost-effectiveness of adding remote monitoring eyewear to the fNMR pathway.

Ethical Considerations

Ethics approval was granted by the Health and Life Sciences Research Ethics Committee, University of Coventry, for the surveys on UK treatment pathways and the Delphi study to evaluate outcomes (P48908). Participants formally consented. Study data were anonymous or deidentified. Participants were not provided with any compensation.

Estimating National Treatment Costs

Medical resource use (primary care, hospital referrals, and inpatient stays) and therapy costs (fNMR and psychological therapy) were estimated per annual cohort. Estimates were based on epidemiological studies, a national survey of UK treatment pathways, and expert opinion [,]. Individual medical treatments were priced based on national reference costs for the 2020/2021 financial year; any earlier prices were inflated using the National Health Service (NHS) Cost Inflation Index pay and prices index [,]. All medical treatments were assumed to occur within 12 months following diagnosis, so no cost discounting was required. Physical and psychological therapy costs were calculated similarly.

Estimating Annual Cohort Long-Term Morbidity Costs

The long-term morbidity associated with unresolved cases was estimated based on the annual number of UK cases, predicted recovery profiles, and reported reduction in health-related quality of life (HRQoL) for any unresolved cases []. For the subgroup who recover fully, HRQoL was assumed to only be affected for a period of 6 months, after which the individual returns to full health (for their age) []. For the subgroup with incomplete recovery, it was assumed that an individual remains in this state for the remainder of their life. The resulting reduction in HRQoL was converted to a monetary value by applying the National Institute for Health and Care Excellence (NICE) value threshold of £20,000 to £30,000 per quality-adjusted life-year []. A currency exchange rate of £1.36=US $1 is applicable.

Estimating Societal Costs

It is recognized that patients with facial palsy with incomplete recovery may cease employment or move away from public-facing roles []. This can lead to reduced economic activity and increased social security benefit payments []. An approximation of these costs was based on data for people of a similar age with disabilities [].

Delphi Method: Valuation of House-Brackmann and Trial Outcome Measures

Trials of fNMR therapy generally report outcomes in terms of a nerve grading system, most frequently the House-Brackmann (HB) grade, rather than a utility measure (eg, EQ-5D) []. A Delphi exercise was undertaken over the period December 2018 to March 2019 to identify the monetary value placed on outcomes reported as an improvement in HB grade, building on earlier studies [,]. Purposive sampling was used to identify national experts in facial palsy and assistive technology. Invitations were sent by email, and once a panel member had consented, they were sent the round 1 Delphi questionnaire (). The questionnaire was piloted for comprehension before use. In the first round, panel members were asked to place a monetary value on recovery from different grades of unilateral facial paralysis, set within monetary ranges recorded in an earlier study undertaken in the United States []. Consensus was set ex ante at 66% agreement. Round 1 responses were analyzed, and any statement showing consensus was extracted. In addition, panel members were provided with a list of other outcomes commonly reported in clinical trials and asked to rate their importance. Round 2 questionnaires were personalized for each panel member by presenting their round 1 scores in the context of overall panel responses; individuals could revise their ratings if they wished. The level of agreement was measured based on the modal consensus approach [].

Modeling of the Cost-Effectiveness of Smart Glasses Added to the fNMR Pathway

A decision analytic modeling approach was used to estimate the impact on costs and health outcomes of adding remote monitoring eyewear to the fNMR pathway []. The economic model was developed in Microsoft Excel and populated with pathway probabilities and health outcomes, as well as costs. The status quo was assumed to be represented by the NHS treatment pathway recorded in a UK survey []. It was assumed that the likelihood of patients being referred for fNMR would not be affected by the addition of telerehabilitation. The economic analysis adopted an NHS perspective and excluded patient-borne and societal costs. The cost of specialist fNMR therapy was based on a typical set of activities (units) identified by experts (physiotherapist and surgeon) and priced based on locally available data. Other health care resource items in the pathway were priced based on NHS national reference costs []. The cost of Frame eyewear was linked to the price identified as acceptable to UK health care providers in independent market research []. The average cost per pathway was estimated by multiplying each resource item by the percentage of patients referred to produce a total cost [-]. Prices were inflated to the 2020/2021 financial year where necessary. Health care costs in the two arms of the model were compared to differences in health outcomes (ie, improvements in HB grade). Changes in HB grade were assumed to remain the same across both arms following therapy, so that any variation in expected health outcomes results from pathway probabilities. The effect size associated with remote monitoring eyewear was based on the reported impact of telerehabilitation in other neurological conditions such as stroke and expert opinion []. The economic analysis followed international standards for health economics research and reporting results [,].

Results

National Treatment Costs

A breakdown of medical treatment and therapy costs is shown in . The total direct health care cost for the cohort of patients diagnosed in 2020/2021 is estimated to be £86.98 million. The majority of this figure (£86.34 million) is linked to medical costs, with 92.9% (£80.239 million/£86.339 million) associated with hospital inpatient stays, 2.7% (£2.335 million/£86.339 million) with outpatient referrals, and 4.4% (£3.765 million/£86.339 million) with general practitioner consultations and primary care prescribing. Overall, physical and psychological intervention costs are much smaller at £0.64 million, with 31.3% (£201,082/£643,292) of the total therapy figure relating to fNMR costs.

Annual Cohort Long-Term Morbidity Costs

The long-term morbidity burden associated with unresolved cases per annual cohort is shown in in terms of quality-adjusted life-years loss. After applying the NICE threshold, this long-term loss is valued at between £238 million and £357 million per cohort. This is 2.7 to 4.1 times the estimated total medical and therapy treatment cost of £86.98 million. The low cost of fNMR therapy in partly reflects the fact that only 4.9% (1120/22,657) of new cases or 17.0% (1120/6570) of unresolved cases are reported to be referred for fNMR.

Societal Costs

Incomplete recovery can lead to reduced economic activity, social security benefit payments, and loss of tax revenues generated through work []. Such societal costs are not included in . However, people of a similar average age (40 y) with 10% to 20% disablement are reported to receive cumulative social security benefit payments of £5000-£10,000 over 5 years []. Applying this to a facial palsy cohort would add a further societal cost of £113.3 million to £226.6 million over 5 years. This produces a final conservative estimate for the total economic burden, including societal costs, of £351 million to £584 million per annual cohort.

Delphi Valuation of HB Grade and Other Trial Outcome Measures

All 26 experts invited to join the Delphi panel accepted (see for panel details). All panel members completed the round 1 questionnaire, with 19 (73.1%) returning the round 2 questionnaire. shows the values placed on a patient’s clinical recovery from different HB grades of paralysis and ratings of the importance of other outcomes reported in trials. The modal rating for high grades of 5 and 6 paralysis achieved predefined consensus (66% agreement) in round 1 with 83% agreement, and for medium severity grades of 3 and 4 paralysis in round 2 with 69% consensus. The modal value for low-severity paralysis failed to show consensus by the end of round 2 (35% agreement). provides a description of HB grades.

also shows the perceived importance of other outcomes most commonly reported in evaluation studies. There was full consensus (100%) on four measures rated as being “very important,” with three further outcomes (psychological distress, patient-borne costs, and NHS treatment costs) achieving a lower level of agreement at 88%‐96%, but well above the ex ante 66% level required for consensus. The one outcome rated “important” (impact on employment) demonstrated consensus at 77% agreement.

Cost-Effectiveness of Smart Glasses Added to fNMR Pathway

A decision tree model was constructed as shown in . An initial node represents patients who do not recover fully in the first 6 months following initial treatment (eg, prednisolone and advice on eye care). From this initial node, there are two branches: the face-to-face fNMR pathway (status quo) and telerehabilitation (Frame pathway). Each branch then develops two arms: one resulting in no further recovery and one showing some or full recovery. There are then three further branches depending on the state of facial paralysis at entry level (ie, HB grades severe, moderate, or mild). The chance of recovery with telerehabilitation, compared to the status quo, was set at the lowest effect sizes reported for telerehabilitation introduced into stroke pathways []. Probability-weighted costs and outcomes were calculated for the two pathways separately, after adjusting corresponding absolute values with joint probabilities of events along the pathway.

presents a bottom-up cost for a full fNMR pathway (status quo). In addition to fNMR appointments, the total cost includes an initial multidisciplinary assessment and administration of botulinum toxin injections. As a check, the final estimate calculated in this way was then compared to contract prices recorded by service commissioners in England and found to fall well within the range reported []. The unit price of the eyewear was set at a high figure of £495.83 (175% of the average £283.33 reported to be an accepted price point in the market survey) [].

presents the incremental cost-effectiveness ratio analysis. This indicates that a full course of fNMR incorporating smart glasses will both cost less and result in better health outcomes than the status quo. In other words, the pathway including Frame eyewear is “dominant.” Per patient, the cost saving is estimated to be £468, and the health gain an improvement of 0.14 in HB grade per person.

Discussion

Principal Results

To our knowledge, this is the first study to assess the economic burden associated with facial palsy, including the costs of treatment and impact on HRQoL, and to model the cost-effectiveness of introducing telerehabilitation into the fNMR pathway. Our economic modeling shows that if remote monitoring (tracking sensors in smart glasses) were to be added to the current pathway, this would prove “dominant” from a health care perspective. In other words, telerehabilitation is predicted to both reduce overall health care costs and result in better outcomes for patients. Scaled up to a national level, the cost saving predicted per patient in the base case would result in savings of up to £3.08 million if all 6570 patients with incomplete recovery in an annual UK cohort were offered digitally supported fNMR, or £0.52 million if it were limited to unresolved cases currently referred for fNMR. Furthermore, according to the Delphi panel valuation, the health gain in terms of HB grade improvement for 6570 patients would represent up to £17.8 million. Within a context where the economic burden associated with residual deficits in unresolved cases is £238 million-£357 million per annual cohort, there is considerable room for investment in technologies that could improve long-term quality of life []. Changes in employment for unresolved cases will increase long-term costs to £1.27 billion [,], offering even more potential if an innovation can reduce societal costs by improving patient outcomes. In the context of a rise in facial palsy cases [-,-], it is even more important to identify and evaluate such innovations.

Limitations

There are a number of limitations to our economic evaluation. First, the inputs used in the decision tree model are based on average costs and outcomes, while studies indicate that treatment and recovery patterns can vary across patients [,]. Ideally, a future model should incorporate such heterogeneity. Second, our economic model used HB grade improvement as the outcome measure since this is most commonly used in studies of effectiveness []. However, we were reliant on data from stroke trials for an estimate of the effect size associated with the introduction of telerehabilitation, so the model could not fully account for different levels of HB grade or severity. Because of this uncertainty, a conservative effect size was used, far lower than the 52% to 85% levels reported for stroke [-]. In terms of cost inputs, the economic model also set a high price for the wearable device at £496 or 175% of the average acceptable price reported in independent market research []. In fact, because incorporating smart glasses reduces health care costs and improves health outcomes, the cost of the device could be increased further up to a break-even price of £964 (US $1234) and still prove cost-effective. The break-even price falls between that of Meta Ray-Ban smart glasses (US $299), which offer artificial intelligence–assisted entertainment [], and the Apple Vision Pro (US $3499), which provides mixed reality experiences []. The unit cost assumed single-use, but Frame devices could be reused by a health care provider, lowering unit costs []. A pathway with digital support might also reduce the number of certain costly treatments associated with fNMR (eg, botulinum toxin appointments). The model made no such assumption. Finally, because our analysis adopted a health care perspective, this means that patient costs are excluded, although some, such as patient-borne travel costs for regular outpatient appointments, can be substantial []. The impact on a person’s income associated with changes in employment for unresolved cases was also excluded but can be significant [,].

In terms of outcomes, our economic model may underestimate benefits. First, it did not consider the possibility of increased adherence to prescribed fNMR exercises, although there is some evidence of this []. Second, it assumed no impact related to accurate monitoring by clinicians, although there is evidence that professionals assume better adherence levels than those recorded by patients themselves []. For stroke, there is international evidence of wide variations in the monitoring of rehabilitation []. Third, telerehabilitation at an earlier point might also result in less entrenched dyskinetic patterns and, in some cases, could minimize more severe residual involuntary movements known as synkinesis []. People with these severe residual deficits experience a greater long-term reduction in quality of life [,-].

In summary, although the data did not allow a full sensitivity analysis to be undertaken, the assumptions made about costs and effectiveness in this preliminary economic model are conservative and probably underestimate real-world benefits and cost savings. Ideally, a lifetime horizon should be explored in future economic analyses. It should also be borne in mind that the model is based on UK fNMR pathways, but these may vary internationally.

Comparison With Other Work

UK guidance on the management of facial palsy currently includes no mention of telerehabilitation or consideration of its cost impact or likely cost-effectiveness []. An evidence gap remains, even though a study of the implications of telemedicine published in 2019 recommended that future research should consider costs []. This study is part of an ongoing research program assessing the value of adding telerehabilitation to the facial palsy fNMR pathway. This includes a market assessment to explore pricing and routes to market [], a national survey of UK treatment pathways [], and a systematic review to update evidence on fNMR effectiveness now added to NICE guidelines on management of facial palsy [,]. Economic evaluations similar to this study are limited. There is no mention of facial palsy in a general review of telerehabilitation [], two systematic reviews of economic analyses of home-based telerehabilitation [], or an economic evaluation of physiotherapy interventions for neurological disorders []. For severe neurological conditions, an evaluation reports that remote physical rehabilitation may cost less and be more effective in the mildest cases []. For stroke, a Cochrane review found evidence of improved outcomes, although limited evidence on cost-effectiveness []. A more recent NICE evidence review has reported that stroke telerehabilitation delivered as an adjunct and telerehabilitation delivered alone are equally effective, but found little research on cost-effectiveness []. An earlier study of VR-based stroke telerehabilitation did report reduced costs (£457 less) but no difference in balance recovery []. Another VR study only reported equipment implementation costs and did not consider wider health care costs or outcomes [].

Conclusions

Long-term morbidity and societal costs associated with facial palsy are estimated to be £351 million to £584 million per annual cohort, indicating significant possible savings if long-term recovery can be improved. Economic modeling confirms that the addition of telerehabilitation to fMNR could improve patient outcomes and reduce costs when compared to current in-person therapy. Because access to specialist neurophysiotherapy services is limited in the United Kingdom, the introduction of telerehabilitation could also enable increased access for currently underserved populations. Further trials with integral economic evaluations in real-world settings are now needed to establish the cost-effectiveness of digitally supported fNMR both early in recovery and in chronic cases. Such studies should meet evidence standards for digital health technologies []. If both clinical (HB grade) and utility (EQ-5D-5L) outcomes are included, this covers the top five outcomes identified by the Delphi panel in this study.

Any future investment in telerehabilitation to support home-based fNMR therapy will need to be driven both by specialist facial therapists and by patients. As with other health care innovations, implementation is likely to meet organizational and cultural barriers [], often reinforced by policy priorities []. Although we have drawn on evidence from stroke telerehabilitation, similar economic analyses are lacking, including for facial palsy following stroke []. However, implementation of telerehabilitation for stroke does highlight certain challenges for policy makers, including potential exclusion of some patients, the need to address staff training, and awareness of the existence of variable practices []. In terms of digital exclusion, there is a need to ensure that implementation does not further exclude patients who require services []. Training in fNMR is essential because a shortage of trained neurophysiotherapists is leading to many general therapists delivering the service; the development of international online specialist training should help address this issue [], together with international guidelines to reduce any variations in practice []. However, further training will be required for the successful introduction of a digital service since interaction online will differ from face-to-face clinical consultations []. For stroke, there are currently no published standards or guidelines for telehealth, and wide variations are reported in quality and monitoring practices []. At a time when 26 of 32 European countries are in favor of implementing home rehabilitation for stroke [], recommendations on the organization of such a service have only recently been published []. Any future implementation of telerehabilitation for facial palsy should be considered in light of international developments for neurological services more widely. Although based on UK data, the findings reported here should be of interest in this international context.

In October, media reports suggested mining giant BHP had accepted a deal to settle about a third of its spot iron ore sales to Chinese customers in China’s own currency, the renminbi (RMB), rather than US dollars.

Those reports still haven’t been officially confirmed, amid ongoing closed-door negotiations between the mining company and China’s state-owned iron ore buyer, China Mineral Resources Group (CMRG).

But headlines quickly jumped to the spectre of “de-dollarisation” and geopolitical turning points.

The reality is less dramatic, but in some ways, more important for Australia.

Changing the invoicing currency doesn’t change how much iron ore China buys. What it changes is who carries the currency risk, which banking systems sit in the middle, and which financial centres earn the fees, deposits and lending business that flow from that trade.

In a new report released today, we find RMB use in Australia is still surprisingly modest. But BHP’s reported deal matters because it exposes how unprepared many Australian banks and firms are for a future where China’s currency plays a much larger role.

The US dollar still dominates

Given China is by far Australia’s largest trading partner, you might expect its currency to loom large in our trade data. It doesn’t.

Australian Bureau of Statistics invoicing data show only a sliver of Australia’s imports and exports are settled in RMB. In the 2023-24 financial year, only 1.4% of merchandise imports by value were invoiced in RMB, and 0.2% of exports.

Across Australia’s total merchandise trade with the world, the Australian dollar (AUD) and US dollar (USD) still dominate.

Even in Australia’s trade with China, RMB settlement has grown only cautiously. It is far more common on the import side (consumer goods and intermediate inputs) than for bulk commodity exports, such as iron ore.

That’s why reports of a deal with BHP drew so much attention. Iron ore is the backbone of Australia’s exports – worth more than A$100 billion a year.

If settling transactions in RMB became the standard for a significant slice of that trade, the flows involved would dwarf today’s RMB usage in Australia’s financial system.

We built the plumbing – but not the capability

You might think we would be well placed to do more business in China’s currency. Over the past decade, Australia has ticked many of the boxes you would associate with becoming an “RMB hub”.

We have a bilateral currency swap line with the People’s Bank of China – meaning our central banks can exchange currencies directly. There’s an official RMB clearing bank in Sydney – offering direct access to China’s onshore RMB and foreign exchange markets.

On paper, there’s also a supportive policy framework. Yet the on-the-ground reality is underwhelming.

As of mid-2025, total Australian investment in assets in onshore Chinese financial markets was about A$40 billion. This is tiny compared with Australian holdings of US securities (around A$180 billion) and still small relative to the scale of our trade with China.

Only a few dozen bonds denominated in RMB have been issued in Hong Kong, with relatively modest amounts outstanding.

Interviews with corporations for our report tell a consistent story. Australian firms that want to borrow, hedge or hold RMB often increasingly find it easier to do so through Chinese banks.

They either transact through the Australian branches of the Chinese banks, or in Hong Kong and Shanghai.

Who gets to cash in?

If more transactions come to be conducted in China’s currency, the most interesting question for Australia is: where does that business land?

If RMB settlements are routed mainly through Chinese banks, then a growing share of the fees, deposits and lending associated with Australia–China trade will sit on their balance sheets, not those of Australian institutions.

Over time, that could erode the role of Australian banks in servicing the country’s largest trading relationship.

There are also implications for regulators. Greater use of RMB in big-ticket exports would deepen Australia’s financial linkages with China’s currency and banking system.

That brings commercial opportunities, but also new channels of vulnerability in a world of sanctions, financial fragmentation and geopolitical tension.

Balancing opportunities and risks

BHP is unlikely to be the last major exporter to consider RMB settlement. As Chinese manufacturers, electric vehicle makers and renewable energy companies expand their presence in Australia, more firms will have both revenues and costs tied, directly or indirectly, to China and its currency.

For Australian banks, the RMB needs to be treated less as an exotic add-on and more as a core capability, alongside the US dollar and the euro. Otherwise, Australian corporations will keep bypassing them in favour of Chinese banks.

For the Australian government, the task is to join up trade and financial policy. If Canberra is serious about both diversifying trade and stabilising relations with China, then RMB usage cannot be left entirely to foreign banks and overseas markets.

For businesses, the RMB is above all a practical tool. It can reduce currency mismatch when both customers and suppliers are in China, and sometimes improve commercial terms.

But it also comes with political and financial stability risks that need to be understood, stress-tested and managed.

OBJECTIVE

To accelerate the decarbonization of China’s road transport sector by supporting the faster adoption of electric vehicles (EVs) and enhancing the associated infrastructure.

DESCRIPTION

The Project entails AIIB providing an A loan of up to United States Dollar (USD) 125 million equivalent in Chinese Yuan (CNY), complemented by a C loan up to USD125 million equivalent in CNY to be mobilized by AIIB on a best-effort basis, to Ping An International Financial Leasing Co., Ltd. (PAIFL) to support its financial leasing services for urban transport electrification in China.

The loan proceeds will support eligible subprojects through lease financing, targeting underserved segments of China’s EV ecosystem. Approximately 80 percent of the proceeds will be allocated to electric light-duty and heavy-duty trucks, as well as electric passenger vehicles in tier three and tier four cities. The remaining 20 percent will be dedicated to charging infrastructure, with a focus on charging stations for electric heavy-duty trucks, charging networks along highways and major roads, and public fast chargers across China. AIIB financing will follow PAIFL’s Sustainable Development Financing Framework, which is aligned with the Green Loan Principles and Social Loan Principles of the Loan Market Association.

ENVIRONMENTAL AND SOCIAL INFORMATION

Applicable Policy and Categorization: AIIB’s Environmental and Social Framework (ESF), including the Environmental and Social Standards (ESS) and the Environmental and Social Exclusion List is applicable to this Project. The Project is placed in Category FI and is expected to have limited adverse environmental and social (ES) impacts. Subprojects classified as Category A or Higher Risk Activities as per AIIB’s ESF will be excluded from this Project. 

Environment and Social Instruments: To manage ES impacts and in accordance with the applicable national laws and regulations and AIIB’s ESF, PAIFL has established an Environmental and Social Management System (ESMS), which shall be enhanced to align with AIIB’s ESF. PAIFL’s enhanced ESMS will exclude all Higher-Risk Activities, consistent with the ESF. Further, according to the ESMS, clients or subprojects with significant non-compliance with environmental, labor practices, health, and safety performance will not be eligible for lease financing by PAIFL.   

Environmental and Social Aspects: The operation of EVs and charging infrastructure is considered clean from an environmental perspective. However, a key environmental concern is the disposal of batteries and E-waste. In the context of leasing finance by PAIFL to retail and commercial customers, responsibility for appropriate disposal of discarded batteries lies with EV manufacturers, as guided by the regulations. EV manufacturers are legally required by the Government to take responsibility for end-of-life vehicle batteries and set up systems for collection, storage, and transferring to recycling firms. In addition, it is required that the disposal of EV chargers or charging piles must be carried out by licensed enterprises that meet specific technical, environmental protection, and occupational health and safety (OHS) standards. During decommissioning, the charging station operator engages licensed enterprises to undertake dismantling and material recovery activities. The social risks are expected to be limited to consumer protection for the retail portfolio, and labor and working conditions, gender, health and safety related risks in the business portfolio. Land ownership and/or land lease agreements of the business portfolio are verified by the business department. To align with AIIB’s strategic focus on inclusive and sustainable development, the Project has supported PAIFL in developing a gender action plan (GAP) aimed at enhancing gender equality at both the operational and institutional levels. 

Occupational Health and Safety (OHS), Labor and Employment Conditions: OHS risks are expected to be limited to vehicle safety and fire hazards resulting from manufacturing defects and poor maintenance. PAIFL’s customers (individual users and commercial entities) are responsible for ensuring timely maintenance of their EVs. Customers receive warranty documentation and guidance directly from auto dealerships, which outline safe usage practices and maintenance expectations. For charging infrastructure, the PAIFL team visits locations to review suitability. However, maintaining safe working conditions and providing fire extinguishers are the responsibility of PAIFL’s customers. PAIFL urges customers to maintain safe working conditions by implementing necessary safety measures. 

Stakeholder Engagement, Consultation and Information Disclosure: PAIFL identifies investors, regulators, and customers as its stakeholders and regularly engages and consults with them to improve its ES risk management practices. The enhanced ESMS will also address relevant stakeholder engagement activities. PAIFL has agreed to disclose an overview of the enhanced ESMS timely on their website. 

Project Grievance Redress Mechanism (GRM) and the Arrangement of Monitoring and Reporting: PAIFL has an established external communications mechanism, as project level GRM, to address ES concerns of individuals, enterprises, and other stakeholders. Ping An Insurance (Group) Company of China, Ltd. (Group) has set up a whistleblowing hotline and email address to receive non-consumer customer service-related complaints from internal and external parties. In addition to these channels, affected persons may also lodge a grievance with the local government hotline–12345. PAIFL and its Group provide an online platform for employees to lodge grievances and provide feedback. The information of the established GRMs and Bank’s Project-affected People’s Mechanism (PPM) will be timely disclosed in an appropriate manner. PAIFL will monitor and report material incidents, accidents, negative public opinion, and lawsuits. PAIFL will submit to AIIB annual ESMS performance reports using an agreed-upon template.

Introduction

The diagnosis and classification of esophageal motility disorders have undergone evolution since the introduction of high-resolution esophageal manometry (HRM) in the early 2000s []. This technological advancement, characterized by closely spaced pressure sensors providing spatiotemporal pressure topography displays, has altered our understanding of esophageal physiology and pathophysiology [,]. The subsequent development and iterative refinement of the Chicago Classification, now in its fourth version, has established a standardized framework for HRM interpretation that has become the global standard for esophageal motility assessment [,]. Despite these advances, significant challenges persist in clinical practice, including substantial interobserver variability even among expert interpreters, time-intensive analysis requirements, and the need for extensive training to achieve competency in HRM interpretation [,].

In recent years, interest in applying artificial intelligence (AI) to medical data has surged [,]. AI in medicine encompasses methods ranging from classical statistical models to advanced deep learning and even generative models. These approaches can rapidly analyze large datasets and automatically extract complex features, making them well-suited to assist in health care data interpretation []. Gastroenterology has seen rapid exploration of AI for endoscopic image analysis, pathology slide interpretation, and other tasks []. Recent comprehensive reviews have demonstrated AI’s expanding role across gastroenterological applications, from polyp detection to diagnostic decision support systems, with particular promise in image-based diagnostics []. Large language models have also emerged as potential tools for clinical documentation and patient education in gastroenterology, though their role in technical interpretation remains under investigation []. Within the field of neurogastroenterology and motility, AI technologies offer particularly compelling advantages given the pattern-based nature of HRM interpretation and the quantitative parameters inherent to manometric analysis. Machine learning algorithms excel at pattern recognition tasks, potentially surpassing human capabilities in identifying subtle abnormalities and maintaining consistent diagnostic criteria application [,]. Furthermore, AI systems can process vast quantities of data instantaneously, enabling real-time interpretation that could transform clinical workflow efficiency [,]. Recent reviews have examined AI applications in general gastroenterology [-]. However, a focused analysis of HRM-specific applications remains lacking.

The evolution of AI methodologies in medical imaging and signal processing has particular relevance to HRM analysis []. Early applications relied on traditional machine learning approaches such as support vector machines and random forests, which required manual feature extraction and engineering [,]. These methods, while showing promise, were limited by their dependence on predefined features and inability to capture complex spatiotemporal patterns inherent to esophageal pressure topography. The advent of deep learning, particularly convolutional neural networks (CNNs), has revolutionized medical image analysis by enabling automatic feature learning directly from raw data [,]. For HRM, this capability allows AI systems to identify novel patterns and relationships that may not be apparent to human observers or captured by traditional metrics. Recent systematic assessments of AI tools in esophageal dysmotility diagnosis have documented the progression from basic automation of landmark identification to sophisticated deep learning models capable of comprehensive Chicago Classification diagnosis []. Contemporary applications now encompass not only HRM but also impedance-pH monitoring, demonstrating the broadening scope of AI in esophageal diagnostics [].

Recent technological advances have further expanded the potential applications of AI in esophageal motility assessment. The integration of complementary diagnostic modalities, such as Functional Luminal Imaging Probe (FLIP) technology and high-resolution impedance manometry, provides multidimensional data that can enhance diagnostic accuracy []. AI platforms have demonstrated 89% accuracy in automated interpretation of FLIP Panometry studies, validating the feasibility of automated esophageal motility classification during endoscopy []. AI systems are uniquely positioned to synthesize these complex, multimodal datasets, potentially revealing pathophysiological insights that single-modality assessment cannot provide []. Moreover, the development of cloud-based computing infrastructure and edge computing capabilities enables the deployment of sophisticated AI models in diverse clinical settings, from tertiary referral centers to community practices [,]. The emergence of generative artificial intelligence and large language model–assisted development has further accelerated model creation, with recent studies demonstrating the successful implementation of Gemini-assisted (Google LLC) deep learning for automated HRM diagnosis, achieving high diagnostic precision across multiple motility disorder categories [].

Despite these promising developments, no comprehensive systematic review has evaluated the full spectrum of AI applications in HRM interpretation or assessed their methodological quality. Therefore, this systematic review aims to (1) systematically evaluate current AI applications in HRM interpretation, (2) assess diagnostic accuracy across different AI methodologies, (3) evaluate methodological quality, and (4) identify barriers to clinical implementation and future research priorities.

Methods

Study Design

The protocol was registered in PROSPERO (International Prospective Register of Systematic Review; CRD420251154237) before initiating the search. This systematic review followed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 reporting guidelines [] (), PRISMA-Diagnostic Test Accuracy () checklist [], and PRISMA-S (Preferred Reporting Items for Systematic Reviews and Meta-Analyses-Search, an extension to the PRISMA statement for reporting literature searches in systematic reviews; ) checklist [].

Database and Searching Strategy

We searched PubMed/MEDLINE, Embase, Cochrane Library, and Web of Science through September 2025, for studies using AI or machine learning to interpret esophageal HRM. Search strategies incorporated keywords and indexed terms, including (“artificial intelligence” OR “machine learning” OR “deep learning” OR “neural network” OR “computer-aided diagnosis”) AND (“high-resolution manometry” OR “HRM” OR “esophageal manometry” OR “esophageal motility” OR “Chicago Classification”; ). Gray literature sources were searched to reduce publication bias.

Additional information sources were systematically searched to identify gray literature and unpublished studies. We searched the medRxiv preprint server [] using the same search terms to identify studies not yet formally published (advanced searching tab). ClinicalTrials.gov [] was searched to identify ongoing or completed trials that may not have been published. Reference lists of all included studies and relevant systematic reviews were manually screened to identify additional eligible studies. No citation reference searches were performed using citation databases.

The search strategy was peer reviewed by information scientists who have extensive expertise in systematic review methodology and database search strategies.

The results from all database searches were exported and deduplicated using EndNote X20 (Clarivate Analytics, 2020). Automated deduplication was performed using EndNote’s duplicate identification algorithm, followed by manual review to identify and remove any remaining duplicates based on title, author, year, and journal. Two reviewers (CSB and EJG) independently screened studies, and discrepancies were resolved by discussion ().

Inclusion and Exclusion Criteria

We included both prospective and retrospective studies that applied an AI-based algorithm to HRM measurements for diagnosing or classifying esophageal motility disorders (eg, achalasia subtypes, esophagogastric junction outflow obstruction, distal esophageal spasm, hypercontractile esophagus, ineffective motility, etc). We excluded nonhuman studies, conference abstracts without full text, studies focusing on anorectal manometry, and studies on other modalities (such as FLIP or pH-impedance) unless they directly involved HRM data integration.

The detailed inclusion criteria are as follows: (1) original research applying AI, machine learning, or deep learning techniques to HRM data; (2) evaluation of diagnostic accuracy, classification performance, or clinical outcomes; (3) inclusion of human participants or HRM studies; and (4) provision of quantitative performance metrics. The exclusion criteria are as follows: (1) review papers, editorials, or case reports without original data; (2) used only conventional manometry without high-resolution capabilities; (3) applied AI exclusively to other esophageal diagnostic modalities without HRM integration; and (4) lacked sufficient methodological detail for quality assessment.

Data Extraction

Two independent reviewers (CSB and EJG) systematically extracted data using a standardized, prepiloted form. Extracted variables included: study characteristics (authors, year, country, and design), patient demographics (sample size, age, and sex distribution), HRM technical specifications (equipment, protocol, and Chicago Classification version), AI methodology (algorithm type, architecture, and training approach), dataset characteristics (size, split ratios, and validation method), performance metrics (sensitivity, specificity, accuracy, and area under the receiver operating characteristic curve [AUROC]), clinical outcomes when available, and implementation considerations. Discrepancies were resolved through consensus or third reviewer (GHB) arbitration. Authors were contacted for missing or unclear data, with a maximum of 3 contact attempts over 4 weeks.

Study Outcomes

Primary outcome measures included diagnostic accuracy metrics for AI systems compared to expert interpretation as the reference standard. Sensitivity, specificity, positive and negative predictive values, and accuracy were calculated when raw data were available. For studies reporting only AUROC values, these were extracted directly. Meta-analysis was planned if sufficient homogeneity existed across studies; however, due to significant heterogeneity in AI approaches, patient populations, and outcome definitions, a narrative synthesis was performed.

Secondary outcomes included: external validation performance compared to internal validation, processing time for automated interpretation, comparison with trainee interpretation, interrater reliability metrics, and clinical outcomes when reported. Subgroup analyses examined performance differences by: AI methodology (traditional machine learning vs deep learning), disorder category according to the Chicago Classification, validation approach (internal vs external), and year of publication to assess temporal trends.

Quality Assessment

We assessed the methodological quality and risk of bias of each included study using the QUADAS-2 (Quality Assessment of Diagnostic Accuracy Studies-2) tool. This tool evaluates risk of bias in 4 domains: patient selection, index test, reference standard, and flow and timing. For each domain, we judged the risk of bias as low, high, or unclear based on the information reported in the study, and we also noted any concerns regarding applicability to the review question []. Two reviewers (CSB and EJG) performed the QUADAS-2 assessments independently, with disagreements resolved through discussion.

Results

Study Selection and Inclusion

Literature search yielded 411 studies from databases and 1 additional record from manual screening. After removing duplicates, 175 studies remained. Following title and abstract screening, 100 full-text papers were assessed for eligibility. Of these, 83 were excluded. Ultimately, 17 studies met inclusion criteria (Figure 1).

is the PRISMA flow diagram for systematic review of AI applications in HRM (2013-2025). Literature search across PubMed/MEDLINE, Embase, Cochrane Library, and Web of Science (database inception through November 2025) identified studies applying AI, machine learning, or deep learning techniques to interpret HRM for diagnosis of esophageal motility disorders. The diagram illustrates the screening process.

Study Characteristics

Studies were published between 2013-2025, with 82% (14/17) of the studies published in 2020 or later. The studies with clearly documented patient numbers included: Hoffman et al [], with 30 participants with dysphagia, Rohof et al [], 50 patients with gastroesophageal reflux disease, Jungheim et al [] with 15 healthy volunteers, Kou et al [] with 2161 HRM cases, Kou et al [] study with 1741 HRM cases, Wang et al [] with 229 esophageal motility cases from 229 individuals, Surdea-Blaga et al [] with 192 HRM studies (patients), Rafieivand et al [] with 67 patients, Zifan et al [] with 60 patients, and Lankarani et al [] with 43 patients. The total confirmed patient count from studies with explicit numbers was at least 4588 patients, though several studies did not report exact patient numbers. Publication years ranged from 2013 to 2025, with 82% (14/17) published after 2020, reflecting the recent emergence of this field. Study designs were predominantly retrospective cohort studies (n=15, 88%), with 2 methodological development studies (n=2, 12%; Rohof et al [] and Kou et al []). No prospective validation studies were identified. All studies used the Chicago Classification as the reference standard, with varying versions used across studies ().

Time Trend of AI Application in HRM Interpretation

The application of AI to HRM interpretation has shown continuous evolution since 2013. Early pioneers such as Hoffman et al (2013) [] applied artificial neural networks to pharyngeal HRM classification, achieving 86.5%-94% accuracy with 335 swallows. During this initial period (2013-2016), researchers focused primarily on automating specific parameter measurements. Rohof et al (2014) [] created the automated impedance manometry analysis automated analysis system with excellent reproducibility (intraclass correlation coefficient: 0.94-0.95), and Jungheim et al (2016) [] applied machine learning to calculate upper esophageal sphincter restitution times.

A methodological shift occurred around 2018 when researchers began adopting deep learning approaches. Jell et al (2020) [] achieved 97.7% accuracy in automated swallow detection using supervised machine learning. The period from 2020-2022 saw widespread adoption of CNNs. Czako et al (2021) [] achieved 97% accuracy for integrated relaxation pressure (IRP) classification using InceptionV3 (Google LLC) CNN with 2437 images. Kou et al (2021) [] developed both an unsupervised variational autoencoder analyzing 32,415 swallows from 2161 patients and a supervised long short-term memory network achieving 83% accuracy []. Wang et al (2021) [] implemented temporal modeling with Bidirectional Convolutional long short-term memory networks, reaching 91.32% overall accuracy. Romanian researchers, including Surdea-Blaga et al (2022) [] and Popa et al (2022) [], achieved 86% and 94% accuracy, respectively, for Chicago Classification automation.

Recent studies from 2023 onwards have explored increasingly sophisticated and diverse approaches. Zifan et al (2023) [] used shallow machine learning approaches, including logistic regression, random forests, and k-nearest neighbors, to analyze distension-contraction patterns in 60 patients with functional dysphagia, achieving 91.7% accuracy with logistic regression for proximal segments and 90.5% with random forests for distal segments. Rafieivand et al (2023) [] developed a fuzzy framework with graphical neural network interpretation, achieving 78% single-swallow accuracy but 92.54% patient-level accuracy in 67 patients. Zifan et al (2024) [] further refined their approach using support vector machines to analyze distension-contraction plots, achieving an AUROC of 0.95 in 60 patients. Lankarani et al (2024) [] pioneered noninvasive acoustic analysis combined with AI, achieving 97% accuracy for IRP prediction in 43 patients. Most recently, studies have incorporated large language models, with Popa et al (2024) [] integrating Gemini with deep learning, while Wu et al [] (2025) developed mixed attention ensemble approaches ().

Diagnostic Accuracy Across Studies

Overall diagnostic accuracies ranged from 78% to 97% across the 17 included studies. The highest accuracies were achieved for specific applications: IRP classification (97%) [], acoustic IRP prediction (97%) [], and swallow detection (97.7%) []. For Chicago Classification automation, accuracy varied from 86% to >93% [,]. Functional dysphagia studies demonstrated segment-specific performance differences, with Rafieivand et al [] highlighting the importance of patient-level versus swallow-level accuracy (92.54% vs 78%).

Notably, none of the studies provided detailed performance metrics for individual Chicago Classification categories, such as achalasia subtypes or specific motility disorders. This absence of disorder-specific sensitivity and specificity data limits understanding of AI performance across the full spectrum of esophageal pathology and represents a critical gap for clinical implementation ().

Methodological Quality

QUADAS-2 assessment revealed variable methodological quality across the 17 included studies (). For the patient selection domain, no studies demonstrated low risk of bias, with 14 (82%) studies showing unclear risk primarily due to unreported sampling methods, and 3 (18%) studies showing high risk: Hoffman et al [] included only disordered cohorts without healthy controls, Jungheim et al [] tested only healthy volunteers limiting representativeness, and Lankarani et al [] had a small specialized cohort.

The index test domain showed the strongest methodological rigor, with 15 (88%) studies demonstrating low risk of bias through appropriate model training and validation separation. Only 2 (12%) studies showed unclear risk: Rohof et al [] due to calibration on the same dataset raising overfitting concerns, and Jungheim et al [] due to the small sample size (n=15), creating uncertainty in algorithm performance.

For the reference standard domain, 14 (82%) studies had a low risk of bias using expert-determined Chicago Classification labels. Further, 3 (18%) studies showed unclear risk: Rohof et al [] focused on automated metric agreement rather than diagnostic ground truth, and both studies by Zifan et al [,] had limited details on reference adjudication. One study by Kou et al [] showed a high risk as it lacked a validated reference standard for unsupervised clusters.

Flow and timing assessment revealed low risk in 15 (88%) studies, with all patient data included in analyses. One study showed unclear risk (Czako et al []) due to the exclusion of 8 patients with probe-placement failure, and 1 study (Popa et al []) demonstrated high risk by excluding indeterminate cases from analysis, introducing potential spectrum bias.

The predominance of unclear risk in patient selection highlights a systematic reporting deficiency across the literature, with most studies failing to document recruitment and enrollment methods adequately. This pattern, combined with the complete absence of external validation noted elsewhere, raises concerns about the generalizability and real-world applicability of these AI systems.

Secondary Findings

None of the 17 included studies performed external validation using datasets from different institutions or periods. All studies relied on internal validation methods, including train-test splits, k-fold cross-validation, or other internal validation approaches. This complete absence of external validation represents a critical limitation in assessing the generalizability of AI models for HRM interpretation. Studies using k-fold cross-validation [,,,,] reported more conservative performance estimates compared to simple train-test splits, suggesting potential overfitting in single-split validation approaches.

Discussion

Principal Findings

The systematic synthesis of current evidence reveals that AI applications in HRM have demonstrated strong technical performance, with diagnostic accuracies ranging from 78% to 97%, while facing substantial translational challenges. The evolution from traditional machine learning algorithms (86.5%-94% accuracy) to deep learning architectures capable of 97% accuracy for specific tasks represents significant technological progress [,,]. These advances occur within the broader context of AI transformation in gastroenterology, where similar trajectories have been observed in colonoscopy, capsule endoscopy, and inflammatory bowel disease assessment, suggesting that the integration of AI into clinical gastroenterology practice is inevitable rather than speculative [,].

The innovation of AI in HRM extends beyond mere automation. These systems represent a major change in how we approach esophageal motility diagnostics [-], offering solutions to important clinical needs: the global shortage of motility experts, the need for rapid and consistent interpretation [], and the potential for telemedicine integration to serve underserved areas [,].

The diagnostic accuracy achieved by current AI systems, particularly for IRP classification and automated Chicago Classification, addresses a fundamental limitation of HRM interpretation: interobserver variability. AI systems maintain consistent diagnostic criteria application while human experts demonstrate significant intraobserver variability on repeated assessments. This consistency could enable more reliable phenotyping of esophageal motility disorders, facilitating precision medicine approaches that move beyond categorical diagnoses to individualized pathophysiological assessment. The superior performance of AI in quantitative parameter calculation eliminates measurement variability that has plagued HRM interpretation since its inception [].

These accuracy levels have important implications for clinical practice. With health care systems facing increasing pressure to reduce costs while improving outcomes, AI-enabled HRM interpretation could decrease repeat procedures and reduce unnecessary testing costs [,]. Moreover, the consistent application of diagnostic criteria could reduce misdiagnosis-related treatment failures that currently affect a considerable number of patients with esophageal motility disorders [,].

However, the apparent success of AI systems must be contextualized within significant methodological limitations identified through quality assessment. Most critically, no studies demonstrated low risk of bias in patient selection, with 82% (14/17) showing unclear risk due to unreported sampling methods and 18% (n=3) showing high risk due to biased cohort selection [,,]. This systematic deficiency in documenting recruitment and enrollment methods raises fundamental questions about the representativeness of training datasets. The complete absence of external validation across all 17 studies compounds these concerns about generalizability. Internal validation consistently overestimates model performance, and the lack of testing on datasets from different institutions, HRM systems, or patient populations means we have no evidence of real-world performance [].

The complete absence of prospective clinical trials represents the most critical barrier to clinical translation. While retrospective studies demonstrate technical feasibility with accuracies of 78%-97%, these controlled environments fail to capture the complexities of real-world clinical practice. Prospective trials are essential to evaluate: (1) how AI systems perform with real-time data acquisition variability, (2) whether AI recommendations alter clinical decision-making, (3) patient outcomes following AI-guided treatment, and (4) integration challenges within existing clinical workflows. Without such evidence, even the most accurate AI models remain research tools rather than clinical instruments [-].

The evolution through distinct phases of AI development in HRM mirrors broader trends in medical AI but also reveals unique challenges specific to esophageal motility assessment. The transition from traditional machine learning to deep learning approaches yielded substantial performance improvements, yet the “black box” nature of deep learning models poses particular challenges in a field where pathophysiological understanding drives therapeutic decision-making []. Clinicians require not just diagnostic labels but mechanistic insights that inform treatment selection between medical therapy, endoscopic intervention, or surgical management. The development of explainable AI models that provide interpretable features and confidence metrics represents a critical priority for clinical acceptance []. Recent advances in attention mechanisms and gradient-based visualization techniques, as demonstrated in the Popa et al [] study using LIME (Local Interpretable Model-Agnostic Explanations), offer promising approaches for making AI decision-making transparent and clinically meaningful.

The integration of multiple diagnostic modalities through AI platforms addresses a longstanding limitation of isolated HRM interpretation. The combination of manometric, impedance, and complementary data provides a more comprehensive assessment of esophageal function than any single modality alone []. AI systems excel at synthesizing these complex, multidimensional datasets, potentially revealing pathophysiological patterns invisible to conventional analysis. The Zifan et al (2023 [] and 2024 []) work on distension-contraction plots illustrates how AI can extract diagnostic value from data presentations that challenge human interpretation. This capability becomes particularly relevant with the Chicago Classification version 4.0 emphasis on provocative testing and positional changes, which generate substantially more data requiring integration and interpretation [].

The absence of disorder-specific performance metrics across all 17 studies severely limits clinical applicability. While overall accuracy appears promising (86%-97%), clinicians need to know how AI performs for specific conditions: distinguishing achalasia subtypes (critical for treatment selection), detecting subtle ineffective esophageal motility (often missed by novices), or identifying rare disorders such as jackhammer esophagus. A system with 95% overall accuracy but poor performance in type II achalasia, for instance, could lead to inappropriate treatment recommendations. Future studies must report sensitivity and specificity for each Chicago Classification category to enable informed clinical decision-making.

Implementation barriers identified across studies reveal a complex interplay of technical, regulatory, clinical, and economic factors. The incompatibility with existing HRM systems reflects the proprietary nature of medical device software and the lack of interoperability standards. The regulatory uncertainty surrounding AI medical devices requires proactive engagement between developers, clinicians, and regulatory agencies to establish appropriate evaluation frameworks [,]. Despite these barriers, the economic rationale for AI implementation is strong. High-volume centers could achieve cost-effectiveness through improved workflow efficiency and reduced need for expert consultation [,,], though specific economic analyses are needed to quantify these benefits. The lack of specific reimbursement codes for AI-assisted interpretation creates financial uncertainty that discourages adoption []. The potential for AI to enable task-shifting from specialists to general gastroenterologists could address workforce shortages and improve access to motility assessment, particularly in underserved areas.

The ethical implications of AI implementation in HRM diagnostic practice deserve careful consideration []. The potential for algorithmic bias, particularly affecting populations underrepresented in training datasets, could exacerbate existing health care disparities. The predominance of studies from North American, European, and select Asian centers raises concerns about applicability to African, Latin American, and other underrepresented populations with different disease phenotypes and genetic backgrounds []. Development of quality assurance programs that monitor AI performance and identify edge cases requiring human review will be essential for maintaining patient safety.

Moving from laboratory validation to clinical implementation requires addressing multiple translational gaps simultaneously. First, prospective multicenter trials must demonstrate that AI systems maintain performance across diverse patient populations, HRM equipment, and clinical settings. Second, health economic analyses must quantify whether efficiency gains justify implementation costs—a critical requirement for hospital administrator buy-in and insurance coverage. Third, regulatory pathways need clarification: Should AI-HRM systems be classified as clinical decision support tools or diagnostic devices? Each classification carries different validation requirements and liability considerations. Finally, implementation science research must address workflow integration, user training requirements, and change management strategies to ensure successful adoption [].

Future priorities must focus on multicenter validation studies, development of explainable AI models, integration with evolving diagnostic frameworks, and systematic addressing of regulatory and economic barriers. The ultimate success of AI in HRM will depend not on technological sophistication alone but on thoughtful integration that preserves clinical judgment while enhancing diagnostic accuracy and efficiency. To achieve clinical translation, the field must transition from technical validation to clinical validation through (1) prospective trials comparing AI-assisted versus standard interpretation on patient outcomes, (2) disorder-specific performance benchmarking across all Chicago Classification categories, (3) cost-effectiveness analyses demonstrating economic value, (4) regulatory sandbox programs allowing controlled real-world testing, and (5) implementation science studies optimizing integration strategies. Until these translational requirements are met, AI in HRM will remain a promising technology awaiting clinical realization.

Study Limitations

This systematic review has several limitations that should be considered when interpreting the findings. First, the heterogeneity in AI methodologies, patient populations, and outcome definitions precluded meta-analysis, limiting our ability to provide pooled estimates of diagnostic accuracy. Second, we excluded non-English language publications, potentially missing relevant studies from non–English speaking countries. Third, the absence of standardized reporting guidelines for AI studies in HRM made quality assessment challenging, particularly regarding technical aspects of model development. Fourth, publication bias could not be formally assessed due to the diversity of study designs. Fifth, the lack of clinical outcome data across all studies prevented assessment of the real-world impact of AI implementation on patient care, treatment decisions, and health care costs. Finally, critical limitations include the complete absence of low-risk patient selection across all studies, the lack of disorder-specific performance metrics for individual Chicago Classification categories, the absence of prospective clinical trials, no cost-effectiveness analyses, and insufficient direct comparisons between AI and human interpreters using standardized metrics. These gaps collectively limit our ability to assess the true clinical utility and implementation readiness of AI systems in HRM interpretation.

Conclusions

This systematic review provides comprehensive evidence that AI applications in HRM have achieved remarkable technical capabilities while facing substantial challenges in clinical translation. The diagnostic accuracies of 78%-97% demonstrate the potential for AI to standardize and enhance HRM interpretation. However, the complete absence of external validation, systematic deficiencies in patient selection documentation, and lack of clinical outcome studies highlight the critical gap between technological capability and clinical utility. Additionally, the limited reporting of patient demographics across included studies—reflecting the methodological focus of AI development papers—represents an ongoing challenge for assessing generalizability across diverse populations. Future AI validation studies should systematically report demographic characteristics, including age, sex, race or ethnicity, and geographic location, to enable evaluation of algorithmic performance across patient subgroups and identify potential disparities in diagnostic accuracy that could affect equitable clinical implementation.

LONDON, Nov 27 (Reuters) – The new head of the Bank for International Settlements has said reining in hedge funds’ ability to make highly leveraged bets in government bond markets should be a key priority for policymakers given rapidly increasing public debt levels.

Pablo Hernández de Cos, who took over as General Manager of the umbrella body for central banks in July, said the combination of high debt and growing role of non-bank financial institutions (NBFIs) such as hedge funds in bond markets posed new financial stability risks.

Sign up here.

The worry is their use of leveraged “relative value” trades like cash-futures basis trades, which look to exploit small price differences between bonds and their futures contracts.

These strategies have boomed in the U.S. and other major economies but have been in the sights of regulators after margin calls on U.S. Treasury future trades in 2021 fuelled a bout of turmoil in the world’s biggest government bond market.

“Around 70% of bilateral repos taken out by hedge funds in U.S. dollars and 50% in bilateral repos in euros are offered at zero haircut, meaning that creditors are not imposing any constraint on leverage using government bonds,” de Cos said in a speech at the London School of Economics.

With ageing populations and rising defence spending projected to push the debt-to-GDP ratio of advanced economies to 170% by 2050 absent fiscal consolidation, de Cos said reining in NBFI leverage was a “key policy priority”.

He called for a “carefully selected combination of tools” but highlighted two specific measures as likely to be particularly effective.

One of those was the greater use of central clearing, so government bond market players are treated more equally. The other was for “minimum haircuts” – or discounts – to be applied to the value of the bonds hedge funds use as collateral, to limit their leveraged plays.

“The growing intermediation of record-high public debt levels by NBFIs introduces significant new financial stability challenges,” de Cos said, adding that haircuts should be applied in a targeted manner.

In the context of these new risks, he said central bank swap lines remained “critical” to stabilise the global financial system at times of acute distress.

Keeping inflation in check will remain the most effective way to support debt sustainability by reducing risk premia, while central bank independence remains vital too.

“Against the backdrop of rapidly deteriorating sovereign creditworthiness, the need for credible monetary policy and central bank independence is stronger than ever,” de Cos said.

Our Standards: The Thomson Reuters Trust Principles., opens new tab

BRASILIA, Nov 27 (Reuters) – Brazil’s economy created a net 85,147 formal jobs in October, data from the Labor Ministry showed on Thursday, below the 105,000 expected by economists in a Reuters poll.

The net figure accounts for 2,271,460 jobs opened and 2,186,313 closed during the period, the ministry said.

Sign up here.

It was the lowest monthly net formal job creation since March, and the lowest for any October since the indicator’s methodology changed in 2020, adjusted data showed.

From January to October, 1,800,650 net jobs were created, down from 2,126,843 in the same period last year, according to the adjusted data.

Our Standards: The Thomson Reuters Trust Principles., opens new tab

Stay informed with free updates

Simply sign up to the Financial services myFT Digest — delivered directly to your inbox.

Deutsche Börse has launched a bid to buy private equity-backed fund platform Allfunds for €5.3bn, as the German exchange group seeks to expand. 

Deutsche Börse said on Thursday it had entered into exclusive discussions with Allfunds regarding a potential acquisition, saying a deal would “reduce fragmentation in the European investment fund industry and create a harmonised business with global reach”.

Amsterdam-listed Allfunds currently has a market capitalisation of €4.9bn and is backed by investors including US private equity firm Hellman & Friedman, which has previously explored a sale of the company. Deutsche Börse’s rival Euronext sought to buy the fund company in 2023 for €5.5bn, but dropped the bid. 

A takeover would pave the way for an exit by Hellman & Friedman, which acquired Allfunds in a €1.8bn deal in 2017 before taking it public in 2021. Between them, the private equity group and Singaporean sovereign wealth investor GIC own about 35 per cent of the company’s shares.

An agreement would be the latest in a string of liquidity events for Hellman & Friedman, which last month pulled off one of the largest European initial public offerings of recent years when it took security services company Verisure public in Stockholm at a €13.7bn valuation.

Deutsche Börse is offering €8.80 per share for Allfunds, a roughly 8 per cent premium to the company’s closing share price on Thursday. Shares in Allfunds closed 22 per cent higher, after Bloomberg earlier reported the potential deal. 

Deutsche Börse runs the Frankfurt stock exchange as well as derivatives exchange Eurex. A deal for Allfunds would mark its biggest acquisition in recent years, after the German group’s purchase of software company SimCorp for €3.9bn in 2023. 

Allfunds helps connect fund management products with investors, charging buyers to access its platform and charging sellers to offer their products such as exchange-traded and mutual funds. It has more than €1.7tn in assets under administration and works with thousands of fund groups.

Deutsche Börse said a deal would expand its fund services arm, and would “significantly benefit” EU equity markets, adding that it strongly believes that “a prospering funds industry [is] essential to the EU’s status as a globally relevant financial centre”. European policymakers have for years tried to encourage more investment in the bloc.

Under the terms of the proposed deal, Deutsche Börse said Allfunds shareholders would also be entitled to dividends of up to €0.20 per Allfunds share for the financial year 2026 and €0.10 per share per quarter in 2027.