As artificial intelligence continues its march into clinical decision-making, it is beginning to intersect with some of the more human and uncertain aspects of medicine — including how long a patient might live.

One such development is FaceAge, a deep learning tool from Harvard Medical School that estimates an individual’s biological rather than chronological age from a facial photograph. In clinical trials involving over 6,000 cancer patients, FaceAge demonstrated a strong ability to predict short-term mortality — even performing comparably to oncologists who had full access to medical histories. The algorithm identifies subtle facial cues related to physiological ageing and aggregates them using a dataset of over 100,000 images. Researchers have proposed it could serve as a “fifth vital sign” to inform triage and end-of-life care planning[1].

While early days for formal clinical adoption, these tools hint at a future in which AI might assist — not replace — expert judgment in complex, high-stakes scenarios.

What This Means for the Medico-Legal World

The notion of using AI to estimate biological resilience introduces thought-provoking implications for personal injury law, insurance disputes, and the evaluation of long-term outcomes.

• Supporting Prognostic Context: In cases where prognosis is uncertain, such tools might offer an additional layer of context — particularly where experts are working with large, incomplete, or conflicting data sets.
• Claim Valuation: If AI-derived indicators suggest significant pre-injury physiological frailty, these may, in future, support or challenge existing assumptions around life expectancy — though human interpretation would remain critical.
• Legal Admissibility and Bias: Courts are not yet ready to rely on these technologies as standalone evidence. Their role, if any, may remain in background analysis unless issues of bias and transparency are addressed.

The Broader AI Mortality Landscape

FaceAge is one example among a suite of emerging tools focused on physiological ageing:

• AI-Powered ECGs: Neural networks trained on ECG data have been shown to estimate a person’s age, with “ECG age gaps” correlating with hidden health risks and long-term mortality[2].
• Retinal Age Gap: Algorithms analysing retinal images can predict age and all-cause mortality risk — particularly non-cardiovascular causes[3].
• Metabolomic Ageing Clocks: King’s College London researchers are building clocks based on routine blood work to estimate biological ageing — offering potential insight into pre-existing vulnerability[4].

 Should Lawyers and Experts Take Notice?

Perhaps not today — but soon.

AI-driven biological ageing tools are not yet admissible in court, and rightly so. But their analytical capabilities are growing fast. As litigation becomes increasingly data-heavy, these tools may begin to surface behind the scenes — in underwriting models, pre-litigation planning, and even expert report preparation.

For medico-legal experts, the challenge will be to understand what these tools can and can’t offer, critically evaluate their limitations, and remain informed as the technology evolves. While human expertise will remain the cornerstone of any assessment, AI may increasingly serve as a background lens — helping experts synthesise data at scale or flag inconsistencies in large bundles.

The age of algorithmic mortality isn’t about replacement. It’s about refinement. The most valuable insights may come not from machines or humans alone — but from the dialogue between the two.

References:

1. Tewari, A. et al. FaceAge predicts clinical outcomes in cancer patients. Harvard Medical School, 2023.
2. Attia, Z.I. et al. Age and sex estimation using ECGs with deep neural networks. Nat Commun, 2019.
3. Zhu, Z. et al. Retinal age gap as a predictive biomarker for mortality. Br J Ophthalmol, 2022.
4. Robinson, O. et al. Metabolomic aging clocks and health outcomes. King’s College London, 2024.