Category: 8. Health

Yuchen Cao and Xiaorui Shen conducted a systematic review of AI models used in studies detecting depression in social media users and found major flaws.

Artificial intelligence models used to detect depression on social media are often biased and methodologically flawed, according to a study led by Northeastern University computer science graduates.

Yuchen Cao and Xiaorui Shen were graduate students at Northeastern’s Seattle campus when they began examining how machine learning and deep learning models were being used in mental health research, particularly following the COVID-19 pandemic. 

Teaming up with peers from several universities, they conducted a systematic review of academic papers using AI to detect depression among social media users. Their findings were published in the Journal of Behavioral Data Science.

“We wanted to see how machine learning or AI or deep learning models were being used for research in this field,” says Cao, now a software engineer at Meta.

Social media platforms like Twitter, Facebook and Reddit offer researchers a trove of user-generated content that reveals emotions, thoughts and mental health patterns. These insights are increasingly being used to train AI tools for detecting signs of depression. But the Northeastern-led review found that many of the underlying models were inadequately tuned and lacked the rigor needed for real-world application.

The team analyzed hundreds of papers and selected 47 relevant studies published after 2010 from databases such as PubMed, IEEE Xplore and Google Scholar. Many of these studies, they found, were authored by experts in medicine or psychology — not computer science — raising concerns about the technical validity of their AI methods.

“Our goal was to explore whether current machine learning models are reliable,” says Shen, also now a software engineer at Meta. “We found that some of the models used were not properly tuned.”

Traditional models such as Support Vector Machines, Decision Trees, Random Forests, eXtreme Gradient Boosting and Logistic Regression were commonly used. Some studies employed deep learning tools like Convolutional Neural Networks, Long Short-Term Memory networks and BERT, a popular language model.

When bacteria infect our bodies, they sometimes form sticky mats of sugars and proteins called biofilms to protect themselves. This viscous layer makes it difficult for antibiotics and immune cells to reach the invading microbes, rendering usual therapies less effective. Researchers, led by Li Zhang at the Chinese University of Hong Kong and Ben Wang at Shenzhen University, demonstrated that magnet-driven, light-activated microrobots can cut through this goo and fight biofilms in the sinuses of animals (Sci. Robot. 2025, DOI: 10.1126/scirobotics.adt0720).

“We saw microrobots as a promising way to physically navigate into these hard-to-reach spaces” and attack bacteria directly, says Zhang in an email.

Other scientists have previously proposed using microrobots, which are smaller than 1 mm, to target and disrupt biofilm formations, either mechanically or by delivering chemicals that kill bacteria. But biofilms in the sinuses present a unique challenge for microrobots because our natural immune response to a sinus infection produces a viscous pus that’s hard to get through.

The researchers got around this sinus buildup problem by designing their bots to stir up the goo. External magnets placed near the sinuses guide the robots to align into chains and form spinning swarms that create a mechanical force to break up both thick sinus fluids and biofilms.

The microrobots themselves have a magnetic core and a shell of copper-doped bismuth oxoiodide (BiOI), a light-sensitive material. When exposed to visible light delivered by an optical fiber guided magnetically into the sinuses, electrons in the BiOI jump to a higher energy level, leaving behind positively charged holes. In this electron-hole pair, the excited electrons can react with oxygen to form superoxide radicals, while the holes react with water to produce hydroxyl radicals—both species are toxic to bacteria.

When the BiOI absorbs light, it also heats up, which further breaks down mucus and biofilms.

In live rabbit sinuses, the robots cut through thick mucus and destroyed bacterial biofilms without damaging healthy tissue. In pig sinus tissue, which is more anatomically like human sinus tissue, the microrobots also destroyed biofilms, with only 3% of bacteria surviving the treatment.

“Any piece [of the system] is not particularly novel, but the combination is certainly an advancement,” says Edward Steager, an expert in microrobots for medical applications at the University of Pennsylvania.

After the treatment, cilia in the sinuses, which are tiny hair-like structures that move mucus, can clear out the microrobots.

The researchers think the treatment, with some modifications, could be expanded to treat biofilms in other parts of the body, like the gastrointestinal or urinary tract.

Zhang plans to conduct larger-scale animal studies with the robots and says the team is also exploring prototype development for human clinical trials.

The Trump administration appears to be expanding RSV vaccinations to some adults starting at age 50, down from 60, following the advice of a recently fired panel of government vaccine advisers.

The decision appears on a Centers for Disease Control and Prevention webpage but as of Wednesday wasn’t on the agency’s official adult immunization schedule.

RSV, or respiratory syncytial virus, typically is a coldlike nuisance, but it can be severe, even life-threatening, for infants and older adults. The CDC recommends vaccination for certain pregnant women and a onetime shot for everyone 75 or older. But people as young as 60 with health problems that increase their risk can also get it.

In April, the CDC’s influential Advisory Committee on Immunization Practices recommended expanding RSV vaccination to high-risk adults as young as 50, too. But the CDC lacks a director to decide whether to adopt that recommendation and Health Secretary Robert F. Kennedy Jr. didn’t immediately act.

Last month, Kennedy fired all 17 members of that panel and handpicked seven replacements that include several vaccine skeptics.

The new panel alarmed doctors’ groups last week by ignoring settled science on a rarely used flu vaccine preservative and by announcing a probe of the children’s vaccine schedule. It didn’t revisit RSV vaccination for older adults.

Kennedy already had taken the unusual step of changing COVID-19 vaccine recommendations without consulting the committee.

On Wednesday, a page on CDC’s website said that on June 25, Kennedy had adopted the ousted panel’s recommendation to expand RSV vaccination to high-risk 50-somethings and it is “now an official recommendation of the CDC.”

That move was first reported by Endpoints News.

___

The Associated Press Health and Science Department receives support from the Howard Hughes Medical Institute’s Department of Science Education and the Robert Wood Johnson Foundation. The AP is solely responsible for all content.

Lauran Neergaard, The Associated Press

Patients hoping to treat weight-related medical conditions need more than just antiobesity medications that are effective, but that also pose challenges for successful use.

That means food counseling should be part of a comprehensive treatment plan for obesity, according to experts who summarized the state of the research on the issue.

A group of 18 researchers came together to publish “Nutritional priorities to support GLP-1 therapy for obesity: a joint Advisory from the American College of Lifestyle Medicine, the American Society for Nutrition, the Obesity Medicine Association, and The Obesity Society.” It came out this month in the organizations’ respective journals.

The 24-page guide acknowledges that for some patients, the new glucagon-like peptide-1 receptor agonist drugs (GLP-1s) amount to weight loss via injection. But treatment success depends on a patient-centered approach involving more than a prescription.

Patients need counseling — and then must integrate into their lives — elements ranging from management of gastrointestinal side effects, to prevention of nutrient deficiencies, to preservation of muscle and bone mass.

“Despite the efficacy and growing utilization of these medications, real-world challenges are increasingly evident,” the paper said.

“All these challenges may be partially mitigated by an evidence-based, structured lifestyle program, particularly around food, when prescribing GLP-1s for obesity,” the advisory said. “However, practical guidance for clinicians to implement such an approach is limited.”

Medical effects

Studies have produced results showing GLP-1s have helped patients reduce their weight. But real-world efficacy generally has been lower, and weight regain is common when patients discontinue use of the medications, the advisory said. Research also has demonstrated clinical benefits for conditions ranging from heart failure to sleep apnea to chronic kidney disease to substance use disorders.

Gastrointestinal irritation is common, but not likely to lead to discontinuation. With a loss of appetite and lower energy intake, patients may develop vitamin and mineral deficiencies and symptoms such as fatigue, hair loss, skin flakiness, muscle weakness, poor wound healing and bruising, the advisory said. Muscle mass and bone density also may decrease.

Financial effects

Adherence to the drugs is relatively low for reasons that are unclear, although cost may be a factor. Based on current list prices, the drugs may cost up to $16,000 a year. Prices are lower with manufacturers’ rebates, dropping to $7,000 to $8,000 annually, and lower still with compounded versions, that still cost up to $3,000 a year.

While patients may experience improved quality of life, the GLP-1s have not provided a long-term financial return for health care overall. Several studies “have found that GLP-1 treatment costs exceed health care savings,” and one found patients using GLP-1 drugs had significantly higher health care costs than patients with obesity, but not using the drugs.

“Considering cost-effectiveness, i.e., health gained per dollar spent, most studies find that GLP-1s, even at currently discounted prices, do not meet accepted thresholds for cost-effective therapy,” the study said.

All those factors indicate something more is needed.

“These high costs, lower adherence in practice, and frequent weight regain after discontinuation, each highlight the importance of complementary nutritional and lifestyle counseling to help maximize overall efficacy and cost-effectiveness,” the advisory said.

Barriers to care

The researchers cited a number of difficulties in current practice:

• Visits with primary care physicians and non-obesity medicine specialists are usually short and centered on acute illness or needs, screening discussions, and medication management.
• Access is limited to lifestyle medicine approaches for obesity and its comorbidities. “For example, the Diabetes Prevention Program is known to reduce the risk of progression to diabetes and is covered by major payers, but has not been meaningfully scaled due to regulatory and implementation barriers,” the advisory said.
• There is no American Medical Association approval of category I Current Procedural Terminology codes for health coaching, so that remains a barrier to reimbursement.
• Private and public payer coverage for medical nutrition therapy for obesity remains limited, preventing broad utilization in practice.

“These pressures, alongside a frequent lack of practitioner education about integrating lifestyle management in medicine, have created a dearth of implemented behavioral and lifestyle counseling, accessible and effective referral programs, and integration into existing care delivery systems,” the authors said.

Getting started

The researchers suggested using a 5As Framework — assess, advise, agree, assist, and arrange — to guide physician-patient interaction toward a successful long-term plan. There also will be at least eight elements or nutritional priorities to support GLP-1 therapy for obesity.

• Initiation of GLP-1 therapy with a patient-centered approach.
• Completion of baseline nutritional assessment and screening.
• Management of gastrointestinal side effects.
• Navigation of dietary preferences and intake.
• Prevention and mitigation of nutrient deficiencies.
• Preservation of muscle and bone mass.
• Maximization of weight reduction efficacy.
• Promotion of other supportive lifestyle measures.

Under “arrange,” the experts noted physicians likely won’t go it alone with patients. Registered dietitian nutritionists, behavioral therapists, social workers and case managers all may be part of a team.

Along with medications, food counseling and a new menu, the authors emphasized the need for patients to take up resistance training and other physical activities. Good sleep, stress management and substance use cessation all will be part of the treatment plan.

More research is needed

The GLP-1 drugs still prompt questions and need more research. The researchers note physicians, other clinicians and patients refer to GLP-1s, but there is no widely accepted terminology to describe the medications. There also is room for research on measuring obesity and adopting definitions and treatments for clinical and preclinical obesity.

“In conclusion, although GLP-1s alone can produce significant weight reduction and related health benefits, several challenges limit its long-term success for individuals and populations,” the advisory said.

“Careful attention to evidence-based nutritional and behavior modification can help mitigate the adverse effects of these challenges,” the authors said. “Thus, all clinicians prescribing GLP-1s for obesity management should establish a thoughtful plan of care that includes thorough nutritional and lifestyle counseling before, during, and after the weight reduction period.”