Background: Serious games have become an alternative support for traditional physical therapy. However, many of these games do not address the emotional needs of patients. People with disabilities often experience emotions such as sadness, frustration, and even anger, which can create a barrier to their rehabilitation treatment. Objective: This review presents a comprehensive overview of technologies, techniques, and methods in affective computing as applied to serious games for physical rehabilitation and identifies key gaps to guide future research in this field. Methods: A scoping review was conducted following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, using the databases PubMed, ScienceDirect, IEEE Xplore, ACM Digital Library, PEDro, Springer, and Google Scholar. Results: The initial search yielded 5293 records, of which 9 papers met the inclusion criteria. Data were systematically extracted from these papers based on predefined research questions. Notably, engagement, tiredness, and pain were the most identified emotions, reported in 4 of 9 (50%) studies. Only 3 studies applied theoretical frameworks for emotion classification. Facial expression analysis and gesture recognition were the most frequently used affective computing techniques, yet only 2 studies implemented adaptive gameplay based on affective feedback. Conclusions: The integration of affective computing into serious games represents a promising approach for detecting affective states in patients undergoing rehabilitation. However, the limited number of primary studies, methodological limitations, and potential selection and reference-standard biases limit the reliability and generalizability of the current findings. Future research should prioritize rigorous multicenter designs, standardized evaluation protocols, and multidisciplinary collaboration. Developing these areas is essential to establishing clinical effectiveness.
<img src="https://jmir-production.s3.us-east-2.amazonaws.com/thumbs/59212b6f73c0a6403d9a3197feaac7f2" />
Prediction of Clinically Significant Depressive Symptoms at 2-Year Follow-Up in Older Adults: Machine Learning Study Using the English Longitudinal Study of Ageing
Background: Depression in older adults is often underdiagnosed due to atypical symptom presentation and generational stigma, leading to delayed intervention. Early identification of individuals at risk of developing elevated depressive symptoms is therefore critical, but traditional approaches show limited predictive accuracy. To date, no study has applied machine learning (ML) models to predict clinically significant depressive symptoms at 2-year follow-up in older adults in the United Kingdom using data from the English Longitudinal Study of Ageing (ELSA). Moreover, the impact of encoding strategies for categorical health care variables has not been examined. Objective: This study aimed to develop and evaluate ML models to predict the clinically significant depressive symptoms at 2-year follow-up in older adults using ELSA data. We further compared ordinal and one-hot encoding strategies across different ML architectures and identified key predictors of depressive symptoms at follow-up. Methods: Data were drawn from 4 consecutive waves of ELSA, including participants aged ≥50 years without significant depressive symptoms at the baseline wave (waves 6‐9). Clinically significant depressive symptoms were defined as 8-item Center for Epidemiologic Studies Depression Scale (CES-D 8) scores of ≥4 at the subsequent wave (waves 7‐10). Over 120 features spanning sociodemographic, psychological, and health-related domains were analyzed. Eight ML models were applied, including tree-based ensembles, deep learning architectures for tabular data, distance-based methods, probabilistic methods, and linear methods. Model performance was assessed using the area under the receiver operating characteristic curve (AUROC) and -score. Model interpretability was examined using Shapley additive explanations (SHAP). Sensitivity analyses assessed the robustness of results across alternative CES-D 8 thresholds (≥3, ≥4, and ≥5) and encoding strategies. Results: Across waves, the best-performing models achieved mean AUROC scores of 0.72‐0.73, with a peak of 0.75 in the highest-performing wave. Ordinal encoding consistently outperformed one-hot encoding across all ML models, yielding improvements in AUROCs and -scores, with the greatest increase in tree-based methods. SHAP consistently identified loneliness, sleep disturbances, and low social engagement as strong predictors of elevated depressive symptoms at follow-up. Sensitivity analyses across CES-D 8 thresholds demonstrated robust feature importance, with AUROCs ranging from 0.67 to 0.82. Traditional ML models (random forest, extreme gradient boosting, and support vector machines) generally achieved higher performance than the deep learning models for this task. Conclusions: Our findings demonstrate the feasibility of predicting clinically significant depressive symptoms at 2-year follow-up in UK older adults, with moderate accuracy. Ordinal encoding demonstrates superior performance for health care datasets with inherently ordered categorical features. The identification of consistent risk factors highlights opportunities for developing targeted clinical screening tools and preventive interventions. This study provides new evidence on depressive symptom prediction in the UK context, leveraging longitudinal data from ELSA, and contributes to advancing digital mental health research for aging populations.
<img src="https://jmir-production.s3.us-east-2.amazonaws.com/thumbs/9a8688496d2b5e1d0eaa69410b4ab2dc" />
The Download: a startup has a solution for AI’s groupthink problem
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology.
LLMs are stuck in a groupthink groove. This startup is trying to get them out.
Open up your chatbot of choice—Claude, ChatGPT, Gemini—and type “Give me a random number between 1 and 10.” You’re going to get 7. Almost always.
That won’t work every time—but if it did for you, you may wonder if I have superpowers. I don’t.
The truth is that most large language models are stuck in a rut. They are far more predictable and far less creative in their responses than you might expect. That’s fine for tasks like coding or research, but groupthink is a problem when you’re brainstorming or planning your next vacation.
The Australian startup Springboards has a solution. It built an LLM called Flint, which has been trained to come up with a wider variety of responses than mainstream LLMs to open-ended questions such as “Where should I go in Europe?”
Meet the company pushing chatbots away from the obvious.
—Will Douglas Heaven
The must-reads
I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.
1 Scientists say they have built a cell from scratch for the first time
Built with lab-made DNA, it can feed, grow, and multiply. (CNN)
+ It brings us closer to creating synthetic life. (Quanta)
+ And is arguably the greatest feat of bioengineering yet. (New Scientist $)
+ But also raises concerns over the dangers of synthetic biology. (NYT $)
+ Mirror organisms could threaten life on Earth. (MIT Technology Review)
2 OpenAI has proposed giving the Trump administration a 5% stake
Talks over a public ownership deal come amid rising political pressure.(FT $)
+ OpenAI also proposed other US AI giants providing a 5% stake. (CNBC)
+ That could include Anthropic, Google, and Meta. (Bloomberg $)
+ President Trump says he wants the public to have a stake in AI. (BBC)
3 Singapore has seized a $42 million mansion tied to Nvidia chip smuggling
It was seized as part of an investigation into alleged illegal trading. (BBC)
+ Days earlier, Supermicro’s Taiwan offices were raided in the probe. (FT $)
4 Anthropic’s Fable 5 is back online
But queries posing security risks may be routed to less powerful models. (Axios)
+ Anthropic restored access yesterday after the US lifted an export ban. (BBC)
+ But the battle over how to tame AI has just begun. (WSJ $)
+ Anthropic has launched a new AI science product. (MIT Technology Review)
5 Meta is building its own cloud infrastructure business
It’s exploring two ways of monetizing AI compute and models. (Bloomberg $)
+ One is selling access to models hosted on Meta’s infrastructure. (CNBC)
+ The other is selling “raw” computing power. (TechCrunch)
6 PlayStation will stop releasing games on discs in 2028
Future PS5 games will be digital-only releases. (Verge)
+ The news comes days after reports that GTA VI will have no disc. (BBC)
+ It’s put a nail in physical media’s coffin. (Wired $)
7 A low-cost Chinese AI model is catching up with US giants on their home turf
Western customers are drawn to GLM-5.2’s cheap but powerful model. (Reuters $)
+ Chinese open-source models are spreading fast. (MIT Technology Review)
8 Google has lost its fight against a record €4.1 billion EU antitrust fine
It was charged in 2018 for using Android to block rivals. (CNBC)
9 The UN has launched an “AI for Good” commission
Salesforce CEO Benioff and Rwandan President Kagame will co-chair it. (Axios)
10 People prefer AI impersonators over politicians
The study’s findings raise alarm bells around potential public deception. (404 Media)
Quote of the day
“If AI overdelivers, it will impact financial stability. If AI underdelivers, it will impact financial stability.”
—Torsten Slok from Apollo Global Management shares common concerns about AI at the European Central Bank’s annual conference, Reuters reports.
One More Thing
America was winning the race to find Martian life. Then China jumped in.
In July 2024, after more than three years on Mars, the Perseverance rover came across a peculiar rocky outcrop. Instead of the usual crystals or sedimentary layers, this one had spots. Those specks were the best hint yet of alien life.
NASA began a new mission to bring the rocks back to Earth to study. But now, just over a year and a half later, the project is on life support. As a result, those oh-so-promising rocks may be stuck out there forever.
This also means that, in the race to find evidence of alien life, America has effectively ceded its pole position to its greatest geopolitical rival: China. Beijing is now moving full steam ahead with its own version of NASA’s mission.
Here’s how the search for Martian life has become a contest between two superpowers.
—Robin George Andrews
We can still have nice things
A place for comfort, fun, and distraction to brighten up your day. (Got any ideas? Drop me a line.)
+ The classic arcade game Crazy Taxi is returning.
+ Thom Yorke’s live set from the Sydney Opera House is a reminder of what an extraordinary performer he is.
+ Peer into 1,000 gloriously illuminated New York apartment windows at night in this generative photography project.
+ The Orion constellation dazzlingly displays every stage of star formation in this image from the James Webb Space Telescope.
Top image credit: Sarah Rogers/MITTR | Photos Getty
Please send gloriously illuminated New York apartments to hi@technologyreview.com.
You can follow me on LinkedIn. Thanks for reading!
—Thomas
EFE-8c4, a polyamine from Elaeagnus multiflora, protects neuronal cells by regulating oxidative stress and apoptotic pathways
Neurological disorders, including depression, cognitive impairment, and neurodegenerative diseases, are closely associated with oxidative stress, apoptotic neuronal loss, and impaired neuronal differentiation. Polyamine derivatives from natural products have emerged as potential neuroprotective agents, although their mechanisms remain incompletely understood. In this study, three polyamine compounds isolated from Elaeagnus multiflora fruit were evaluated in SH-SY5Y cell models of oxidative and glucocorticoid-induced stress. Among these, EFE-8c4 exhibited the most pronounced neuroprotective activity, significantly restoring cell viability under corticosterone-induced stress and attenuating oxidative damage induced by hydrogen peroxide. Mechanistically, EFE-8c4 modulated apoptotic signaling by increasing Bcl-2 expression while suppressing Bax, caspase-8, and p53 activation, thereby restoring the balance between pro- and anti-apoptotic pathways. In addition, EFE-8c4 reduced intracellular reactive oxygen species accumulation and enhanced the expression of PCNA and βIII-tubulin, indicating improved cell survival capacity and neuronal phenotype maintenance. Furthermore, EFE-8c4 partially reduced apoptotic cell populations under corticosterone exposure. Collectively, these findings demonstrate that EFE-8c4 exerts multi-target neuroprotective effects through coordinated regulation of apoptosis, oxidative stress, and neuronal differentiation-related pathways, highlighting its potential as a candidate for the treatment of oxidative stress-associated neurological disorders.
Modulating autonomic nervous system activity with transcutaneous auricular vagus nerve stimulation in Parkinson’s disease: a proof of concept study
BackgroundCardiovascular dysautonomia is a debilitating non-motor symptom of Parkinson’s disease (PD) that limits exercise capacity and neurorehabilitation outcomes. Transcutaneous auricular vagus nerve stimulation (taVNS) is an emerging non-invasive neuromodulatory therapy that modulates cardiovascular activity and could potentially serve as an adjunct to exercise, yet its physiological effects on cardiovascular function in PD remains unexplored.ObjectiveThis proof-of-concept, sham-controlled crossover pilot study (N = 8) investigated the acute effects of taVNS on cardiovascular autonomic activity in idiopathic PD.MethodsParticipants underwent active taVNS (30 Hz, 250 μs, 0.1–4 mA) or sham stimulation (0 mA) during a 15-min resting phase, immediately followed by the Ewing Battery of cardiovascular reflexes. Acute autonomic shifts were phenotyped using continuous heart rate variability (HRV) monitoring.ResultsThis proof-of-concept protocol was feasible, as all participants completed the randomized crossover stimulation visits and autonomic reflex testing without adverse events. Baseline autonomic burden (COMPASS-31) was associated with the magnitude of heart rate response to active stimulation. Immediately following stimulation and during the deep breathing challenge, active taVNS was associated with directionally consistent changes in vagally mediated HRV metrics including RMSSD, pNN50, and HF power, relative to sham.ConclusionContinuous HRV monitoring and autonomic reflex testing appears feasible for characterizing acute autonomic responses to taVNS in PD. Active stimulation was associated with directional changes in vagally mediated HRV metrics during the post-stimulation period and during deep breathing, supporting the biological plausibility of acute autonomic modulation. These preliminary findings justify larger, adequately powered studies designed to determine whether taVNS can reliably modulate cardiovascular autonomic regulation and inform rehabilitation optimization in PD.
Neurochemical-hemodynamic-electrophysiological coupling in the neonatal brain: a multimodal MRS-fMRI-EEG investigation
IntroductionInhibitory and excitatory neurotransmitter levels are linked to fast neuronal oscillations and infra-slow hemodynamic fluctuations, suggesting a shared excitation–inhibition (E/I) regulatory framework across measures. However, these relationships may differ in early development, when both excitatory and inhibitory cortical systems are undergoing substantial functional and structural maturation. Consequently, we hypothesize different functional coupling between neurochemical, electrophysiological, and hemodynamic proxies of E/I signaling in healthy full-term neonates compared to what has been observed in adults.MethodsTwenty-five healthy full-term neonates (mean postmenstrual age at study = 40.1 ± 1.4 weeks) underwent multimodal MRI and electroencephalography (EEG) recordings during natural resting-state to provide proxy measures of neural excitation and inhibition. These included frontal and occipital MRS measures of γ-aminobutyric acid (GABA+) and Glx (glutamate + glutamine) levels, and their ratio; EEG source-reconstructed power spectra decomposed into periodic beta (13–30 Hz) and gamma (30–45 Hz) features (center frequency and peak amplitude), relative to total band power and an aperiodic exponent; and infra-slow fMRI BOLD fluctuations (0.01–0.08 Hz) using amplitude of low-frequency fluctuations (mean and fractional ALFF). Crossmodal relationships were assessed using partial correlations controlling for age.ResultsOccipital GABA+ was negatively correlated with beta relative power (r = −0.64, p = 0.01) and fractional ALFF (r = −0.55, p = 0.048), while mean ALFF was negatively correlated with gamma center frequency (r = −0.99, p = 0.02). These relationships were not observed in the frontal cortex. Instead, frontal Glx positively correlated with beta peak amplitude (r = 0.87, p < 0.01) and negatively correlated with beta (r = −0.78, p = 0.02) and gamma (r = −0.79, p = 0.02) relative power, potentially reflecting the existence of regionally distinct maturational trajectories.DiscussionTogether, these preliminary findings suggest that commonly used neurochemical, oscillatory, and hemodynamic proxy measures of cortical excitatory and inhibitory processes may show only modest correspondence at birth, consistent with ongoing and hierarchal cortical development, leading to complex and asynchronous relationships between these measures.
Developing forensic patient-oriented research guidelines: a rapid review using an integrated knowledge translation approach
This paper reports findings from a rapid literature review that informed new guidelines for conducting patient-oriented research in forensic mental health settings. The project adopted an integrated knowledge translation approach at a mental health hospital in Ontario, Canada, engaging a project team that included current forensic patients, hospital staff, and members of an international community of practice. Sources were identified through nine academic databases and targeted grey literature searches, screened independently by two reviewers and extracted using a structured template guided by an a priori framework developed with patients and staff at a knowledge exchange event. Findings were iteratively refined through a patient advisory group, an implementation study, ethnographic observations, and related integrated knowledge translation activities conducted alongside the review. Together, 31 academic and grey literature sources informed a framework organized around five core dimensions: 1) Resourcing, orientation, and training; 2) Confidentiality, consent, and compensation; 3) Relationships, shared understanding, and support; 4) Levels of engagement; and 5) Evaluation and sustainability. Guided by cross-cutting principles common among participatory mental health research, such as dignity, trust, respect, and a commitment to redressing power and attending to forms of epistemic injustice, the guidelines respond to distinctive constraints of forensic environments while highlighting opportunities to promote authentic co-production and sustain patient involvement in research. Recommendations include dedicated resources and capacity-building for patients; relational, ongoing consent practices co-developed with patients; flexible patient researcher roles with fair, paid compensation; and sustained institutional support for participatory practices. We call on forensic hospitals and secure settings to adapt and evaluate these guidelines and to invest in expanding patient leadership to advance the field.
Age-stratified multimodal MRI and machine learning to explore autism-related brain characteristics in youth
PurposeAutism is a common neurodevelopmental condition (NDC) that is characterized by restricted, repetitive behaviors and social communication differences that can impact the daily functioning of individuals. The clinical diagnosis of autism can be challenging, mainly due to its behavioral variability and frequent co-occurrence with other NDCs. This study investigates the ability of machine learning-based classification models trained using multimodal neuroimaging data combined with feature-importance analyses to identify development-specific brain characteristics associated with autism.ApproachA total of 144 participants aged 5 to 18 years with structural MRI (sMRI), diffusion MRI (dMRI), and resting-state functional MRI (rs-fMRI) data available were obtained from the Autism Brain Imaging Data Exchange (ABIDE) database. Radiomic features were extracted from each MRI data modality and used to train support vector machine (SVM) classifiers to identify neuroimaging patterns associated with autism. Single MRI modality classifiers, as well as one combining all three modalities, were trained for comparison purposes. To investigate age-specific effects, the same approach was followed for three age sub-groups: younger children (5–11 years), adolescents (12–18 years), and the entire 5–18 years age cohort. Model performance was evaluated using leave-one-out cross-validation across 30 diagnosis-balanced data splits. Feature-importance analyses were conducted to identify the most important neuroimaging features for classification.ResultsThe classification accuracies of the unimodal models ranged from 68.3% to 75.3% for sMRI, from 69.3% to 77.6% for dMRI, and from 66.3% to 69.9% for rs-fMRI data across age groups. Among all single imaging modalities and age groups, dMRI showed the highest performance with a 77.6% accuracy in younger children (5–11 years). The multimodal approach improved classification performance when compared to the unimodal models in all age groups, achieving accuracies of 78.9%, 76.7%, and 70.5% in the younger, adolescent, and entire age cohorts, respectively. Our findings indicate that multimodal classifiers integrating complementary structural, microstructural, and functional imaging features result in a more comprehensive representation of brain features that strengthens model performance. The most informative brain regions for classification differed between children and adolescents while several diffusion-derived features significantly correlated with social responsiveness scores, emphasizing the clinical importance of studying white and gray matter microstructure in autism.ConclusionsThis study demonstrates the potential of multimodal neuroimaging-based machine learning models to identify development-specific biomarkers associated with autism. The results highlight the value of integrating age-stratified analyses of multimodal neuroimaging to better capture autism-associated developmental brain characteristics. The framework adopted in this study could be extended to explore other NDCs in the future.
Enhancing psychiatry education: effectiveness of a psychodynamic psychotherapy module for borderline personality disorder for psychiatry residents
BackgroundPsychodynamic psychotherapy is the treatment of choice for borderline personality disorder (BPD); however, psychiatric residents frequently report difficulty in applying it, partly due to the lack of structured training models. This study developed and evaluated the effectiveness of psychodynamic psychotherapy learning modules for BPD among Indonesian psychiatry residents.MethodsA quasi-experimental pre-/post-test control group study using mixed methods was conducted across nine psychiatric residency programs in Indonesia. Thirty-four residents were recruited, of whom 33 completed the study. Learning outcomes were assessed using multiple-choice questions and the Psychodynamic Formulation Competency Assessment Scale (PF-CAS) and Practical Competency Assessment Scale (PC-CAS). The module program was evaluated by the participants using the Indonesian version of the Kirkpatrick Level 1 questionnaire.ResultsThe intervention group showed significantly greater improvement in psychodynamic formulation skills (PF-CAS) than the control group (p < 0.001). The multiple-choice scores improved in both groups, with no significant between-group differences. The intervention group showed a numerically greater improvement in Practical Skills (PC-CAS) than the control group, although the difference was not statistically significant. Participants’ feedback was highly positive, emphasizing the usefulness of psychodynamic formulation training, psychotherapy protocols, and supervision.ConclusionImplementation of the psychodynamic psychotherapy for BPD Learning Module enhanced competencies in the cognitive and affective domains and showed promising trends in practical skills. This positive reception highlights its feasibility and potential benefits as part of the residency training curriculum.
Effects of Balint group combined with mindfulness-based stress reduction on humanistic care ability and psychological resilience among obstetric nurses
BackgroundHumanistic care competence and psychological resilience are essential for improving nursing quality, particularly in high-stress specialties such as obstetrics. However, effective interventions that simultaneously enhance both interpersonal and intrapersonal capacities among nurses remain limited.MethodsA total of 87 obstetric nurses from a tertiary hospital in Hebei Province, China, were enrolled and allocated into three groups: a combined Balint group and mindfulness-based stress reduction (MBSR) intervention group, a Balint group, and a control group (n = 29 each). The intervention was conducted over 8 weeks. Outcomes, including humanistic care competence, empathy, emotional intelligence, and psychological resilience, were measured at baseline, post-intervention, and 6-week follow-up using validated Chinese versions of standardized scales. Data were analyzed using repeated-measures analysis.ResultsThe combined intervention group showed significantly greater associations with improvements in all outcomes compared with the Balint and control groups (all P < 0.001). Empathy, humanistic care competence, emotional intelligence, and psychological resilience were significantly higher after the intervention and continued to show positive trends at follow-up. Although the Balint group alone also demonstrated moderate improvements, the combined intervention consistently produced stronger and more sustained associations.ConclusionThe integration of Balint group and MBSR interventions eff is associated with enhanced psychological resilience and humanistic care competence among obstetric nurses. This study builds on previous research by examining the combined effect of reflective and mindfulness-based approaches in a specific clinical population, providing evidence for a feasible strategy to improve nurses’ professional quality and mental well-being.