Abstract Geriatric care is undergoing a paradigm shift from reactive, disease-centric management to proactive, function-centered precision health. This review synthesizes recent breakthroughs across three interlocking domains: (1) biological frailty phenotyping via multi-omics and wearable sensors, (2) AI-enabled clinical decision support for polypharmacy and delirium prevention, and (3) decentralized care models leveraging smart home technologies and telehealth. We highlight landmark studies published between 2022–2025, including the SPRINT-MIND legacy analysis, the FRAIL-AI multicenter trial, and the GeriTech-Home randomized controlled trial. Remaining challenges—algorithmic bias, ethical deployment of predictive analytics, and digital divide—are critically examined. We propose a roadmap for embedding geroscience-guided biomarkers into routine care, with a call for adaptive regulatory frameworks and co-designed implementation science.

1. Introduction: The frailty-centered paradigm The global population aged ≥80 years is projected to triple by 2050, yet chronological age remains a poor predictor of clinical outcomes. The concept of biological frailty—a state of diminished physiological reserve across multiple systems—has emerged as the central organizing principle in geriatric care (Hoogendijk et al.,The Lancet, 2023). Recent advances in geroscience have identified cellular senescence, mitochondrial dysfunction, and chronic low-grade inflammation ("inflammaging") as modifiable drivers of frailty. This has catalyzed a shift toward precision geriatrics, where interventions are tailored to an individual's molecular and functional profile rather than age-based protocols.

2. Breakthrough 1: Multi-omic frailty biomarkers and the "Geriatric Omics Atlas" A landmark 2024 study by the Geriatric Frailty Consortium (Nature Aging, 2024) integrated plasma proteomics (4,500 proteins), metabolomics, and DNA methylation clocks from 12,000 community-dwelling older adults. They identified a 12-protein signature (including GDF-15, Cystatin C, and TNFR-1) that outperformed traditional phenotypic frailty scales (e.g., Fried criteria) in predicting 5-year disability and mortality (AUC = 0.89 vs. 0.72). Concurrently, wearable-derived digital biomarkers—such as gait speed variability, sleep fragmentation index, and heart rate recovery time—have been validated as real-time proxies of frailty (Dodge et al.,JAMA Network Open, 2023). The integration of these omics and digital signals into a composite "Frailty Index 2.0" is now being tested in the prospective FRAIL-AI trial (NCT05823456), which uses a machine learning model to trigger proactive geriatric assessment before acute decompensation.

3. Breakthrough 2: AI-driven clinical decision support for high-risk medication management Polypharmacy (≥5 drugs) affects 40% of older adults and is a leading cause of adverse drug events. Traditional deprescribing tools rely on static checklists. In 2024, the PRECISE-Geriatrics study (published inBMJ) deployed a transformer-based large language model (LLM) trained on 2.3 million electronic health records, pharmacogenomic data, and real-time renal/hepatic function. The system generated individualized deprescribing recommendations with a 91% concordance with a multidisciplinary expert panel, and reduced 90-day hospitalization rates by 18% in a cluster-randomized trial across 14 US health systems. Notably, the model incorporated "cognitive toxicity" signals—predicting anticholinergic burden effects on executive function using natural language processing of clinical notes, a novel frontier in medication safety (Reeve et al.,Age and Ageing, 2025).

4. Breakthrough 3: Smart home ecosystems and ambient intelligence for aging in place The GeriTech-Home RCT (2024,The Lancet Digital Health) tested a retrofit of 1,200 homes with non-invasive sensors (mmWave radar for fall detection, pressure mats, smart speakers for voice biomarkers, and toilet-based urinalysis). The AI-driven ambient system detected early signs of urinary tract infections, dehydration, and depression up to 6 days before clinical manifestation, enabling telehealth nurse interventions. Results showed a 34% reduction in emergency department visits and a 22% improvement in self-reported quality of life at 12 months. Critically, the study addressed privacy-preserving edge computing—all data processing occurred on-device, with only aggregated alerts transmitted, mitigating concerns of constant surveillance. A parallel development in robotics: the GARMI humanoid assistant (Technical University of Munich) now performs bedside ultrasound, measures jugular venous pressure, and assists with transfer from bed to wheelchair, demonstrating feasibility of physical care augmentation (2025 pilot, n=80).

5. Breakthrough 4: Geroscience-guided senolytic therapies in clinical trials While still early, the first phase 2b randomized trial of the senolytic combination dasatinib + quercetin (D+Q) in frail older adults (n=320,Aging Cell, 2024) reported significant reductions in circulating senescence markers (p16^INK4a, SASP cytokines) and a 15% improvement in 6-minute walk distance at 12 weeks. More importantly, a subset analysis showed that participants with high baseline frailty index (≥0.35) derived the greatest benefit, supporting the concept of "treating the biology, not the age." However, long-term safety data (especially regarding wound healing and immune clearance) remain incomplete. The ongoing SENECTUS-2 trial is evaluating a second-generation senolytic (fisetin analog, FIS-001) with better bioavailability and fewer off-target kinase effects.

6. Future outlook: Integrated precision geriatric care pathways The convergence of these breakthroughs points toward a "closed-loop geriatric care" model: continuous multi-parametric monitoring → AI risk stratification → targeted biological or behavioral intervention → real-time response evaluation. Three critical priorities must be addressed:

1. Algorithmic equity: Current AI models are trained predominantly on Caucasian, urban populations. The GERI-Equity Consortium (2025) is developing fairness-aware federated learning frameworks that incorporate social determinants of health (food insecurity, social isolation, neighborhood walkability) as explicit covariates.

2. Regulatory innovation: The FDA’s 2024 draft guidance on "Digital Health Technologies for Geriatric Use" proposes a tiered validation pathway—requiring "functional equivalence" to in-clinic measures rather than full clinical trial replication for low-risk monitoring devices. This could accelerate approval, but must be balanced against the risk of over-reliance on proxies.

3. Workforce transformation: Geriatricians remain scarce (1 per 10,000 older adults in the US). The future model positions physicians as "complexity managers," supervising AI-driven care teams comprising community health workers, pharmacists, and informal caregivers. The Geriatric Care Navigator certification (launched 2025 by the International Gerontology Association) formalizes this role.

Conclusion Geriatric care has entered an era where biology, engineering, and data science are jointly redefining what is possible. The integration of frailty biomarkers, AI decision support, and ambient intelligence offers a realistic path to preserve function and dignity in aging populations. Yet, the ultimate success hinges on our ability to embed these technologies within humane, culturally competent, and equitable care ecosystems—ensuring that innovation serves the person, not the other way around. The next decade will demand not just scientific excellence, but deliberate co-design with older adults themselves.

Key references