Interviews with 12 Montreal clinicians identify the clinical-decision-making complexity and tool limitations they face in managing women's cardiovascular risk, plus the features (efficiency, multilingual design, EMR integration, ease of use) they want from an AI-enabled tool.
FACETS 11, 1-12 · 2026
A survey of family caregivers finds social influence is the strongest predictor of their willingness to use AI-enabled caregiving technology for older adults, ahead of cost or trust concerns.
BMC Geriatrics 26(1), 150 · 2025
A small feasibility study shows a novel HOPE machine-learning model can classify depression in older adults from non-intrusive Wi-Fi motion-sensor data, with sleep-pattern features most predictive.
JMIR Aging 8, e67715 · 2025
A scoping review of 125 studies on socially assistive robots for Alzheimer's patients finds most target daily-life support and cognitive therapy, with cognitive assessment far less explored.
Archives of Gerontology and Geriatrics 123, 105409 · 2024
Of 10 included studies, most AI applications for COVID-19 geriatric care were pilot-stage single-model tools tested in high-income countries with limited representation of women and minority groups.
Archives of Gerontology and Geriatrics 116, 105129 · 2024
Of 20 included studies using motion-tracking data for dementia detection, 40% focused on screening for dementia/cognitive decline, showing promise but with clear research gaps around demographic and cultural representativeness.
Dementia and Geriatric Cognitive Disorders Extra 13(1), 28-38 · 2023
A book chapter examining how AI-powered human-robot communication could assist in the early detection of Alzheimer's disease.
Machine Learning and Artificial Intelligence in Healthcare Systems, pp. 331-348 · 2023
An explainable-AI (SHAP-based) bias-mitigation method reduces sex-related unfairness in a COVID-19 severity prediction model for older adults more effectively than standard fairness techniques, without sacrificing much accuracy.
JAMIA Open 9(1), ooaf171 · 2026
Reviewing 42 studies, the authors find no comprehensive framework yet systematically applies equity, diversity and inclusion (EDI) principles across the AI lifecycle in healthcare, with EDI in implementation and within AI teams especially neglected.
PLOS Digital Health 4(7), e0000941 · 2025
Combining a scoping review with two international workshops of 60+ stakeholders, the authors develop the EDAI framework for embedding equity, diversity and inclusion throughout the AI lifecycle in health and oral health care.
Journal of Medical Internet Research 26, e63356 · 2024
Protocol for a scoping review assessing how equity, diversity and inclusion concepts have (and have not) been integrated into the design, development and implementation of healthcare AI.
BMJ Open 13(9), e072069 · 2023
Conference-abstract precursor to the explainable-AI, bias-mitigated COVID-19 severity model later published as the 2026 JAMIA Open paper above.
The Annals of Family Medicine 21 (Supplement 1) · 2023
Short conference paper arguing for integrating equity, diversity and inclusion throughout the AI lifecycle in health.
13th Augmented Human International Conference, pp. 1-4 · 2022
Combining a scoping review with two national expert panels, the authors build the AIFM-ed curriculum framework — five key elements — for teaching AI to family medicine residents across Canada.
JMIR Medical Education 11(1), e66828 · 2025
Surveying McGill family medicine applicants and interviewers finds broad openness to virtual interviews and to learning AI/VR in residency, even though in-person interviews remain preferred long-term.
BMC Medical Education 24(1), 975 · 2024
Of 21 relevant papers, only two describe an actual AI curriculum framework for medical education — most just report individual programs, revealing a lack of pedagogical grounding in how AI is taught to physicians.
JMIR Medical Education 10(1), e54793 · 2024
GPT-4 answered more than 80% of sample Canadian family-medicine certification exam questions correctly, significantly outperforming GPT-3.5, suggesting promise (with caution) for exam-prep support.
Family Medicine and Community Health 12(Suppl 1), e002626 · 2024
Protocol for the scoping review (published in full in JMIR Medical Education, 2024, above) on AI curriculum frameworks and educational programs for physicians-in-training.
JBI Evidence Synthesis 21(7), 1477-1484 · 2023
A practical guide distilling ten core principles for designing and deploying AI tools that genuinely support, rather than replace, shared decision-making between clinicians and patients.
Oxford Textbook of Shared Decision Making in Healthcare, Ch. 61, p. 309 · 2025
Of only 6 studies meeting inclusion criteria, AI use in shared decision-making is still in its infancy, with little attention to patients' values/preferences, explainability, or end-user involvement in design.
JMIR Medical Informatics 10(8), e36199 · 2022
Testing seven machine-learning models on Quebec health-survey data shows XGBoost and NODE best predict metabolic syndrome risk, with age, perceived health and sex as the top drivers.
BMJ Digital Health & AI 2(1), e000065 · 2026
A systematic review of 16 studies finds EHR-based machine-learning models show promise for catching undiagnosed atrial fibrillation in primary care, but suffer from limited external validation and high risk of bias.
PLOS Digital Health 4(10), e0001009 · 2025
Conference-abstract precursor to the full PLOS Digital Health systematic review (above) on machine-learning detection of atrial fibrillation in primary care EHRs.
The Annals of Family Medicine 22 (Supplement 1) · 2024
A scoping review of 11 studies underpins the new DECIDE-Twin framework, a step-by-step guide for developing and implementing AI-enabled digital twins in clinical decision-making.
IEEE Journal of Biomedical and Health Informatics 29(9), 6332-6341 · 2024
A commentary arguing that, given genuine uncertainty over whether AI can possess consciousness, primary care must prepare now for the possibility of conscious AI systems in diagnosis, shared decision-making and resource management.
Family Medicine and Community Health 12(Suppl 1), e002625 · 2024
Conference-abstract precursor to the full systematic scoping review on AI in community-based primary health care, published in JMIR (2021, below).
The Annals of Family Medicine 20 (Supplement 1) · 2022
A web-based decision support system combining sparse fuzzy cognitive maps with a quantum-learning algorithm outperforms standard machine-learning methods at early detection and severity classification of rheumatoid arthritis from scarce primary-care data.
Mathematics 10(3), 496 · 2022
Of 90 included studies, machine learning, NLP and expert systems dominate AI use in community-based primary care (mostly for diagnosis/detection), but reporting quality and risk-of-bias are highly inconsistent.
Journal of Medical Internet Research 23(9), e29839 · 2021
Of 88 relevant studies, AI applications for adolescent mental health focus overwhelmingly on diagnosis (especially suicide risk and autism) and far less on treatment, monitoring, or prognosis.
JMIR Mental Health 12, e70438 · 2025
A focus group of primary care physicians sees AI as potentially cost-effective and useful for adolescent mental health care, but fears over-reliance eroding clinical competency and wants clear medicolegal guidelines first.
BMC Primary Care 25(1), 215 · 2024
Reviews the current near-absence of shared decision-making policy and training in Iran, and proposes a culturally adapted SDM model built through consensus with Iranian clinicians and SDM experts.
Zeitschrift für Evidenz, Fortbildung und Qualität im Gesundheitswesen 171, 62-67 · 2022
A meta-synthesis of 19 studies finds patient and provider unwillingness plus inadequate infrastructure are the main barriers to patient engagement in safety, while trust-building and patient-centred care facilitate it.
Nursing Forum 56(4), 938-949 · 2021
A half-day shared decision-making workshop for Iranian healthcare professionals raised their intention to use SDM, with training cited as the main facilitator and heavy patient load as the main barrier.
BMC Health Services Research 21(1), 225 · 2021
GCNFusion, a hybrid graph convolutional network combining feature selection and graph embedding, outperforms benchmark methods by 2-5% at identifying influential nodes for information-diffusion prediction across five real-world networks.
Expert Systems with Applications 202, 117053 · 2022
Working with patients, caregivers and providers, the authors co-develop an 8-criteria, weighted decision tool (led by patient safety and risk of harm) to prioritize access to augmentative and alternative communication services in Quebec.
Disability and Rehabilitation: Assistive Technology 17(1), 8-15 · 2022