AI Algorithm Predicts Biological Age from Selfies, Potentially Forecasting Lifespan

Researchers at Mass General Brigham (MGB) have developed a groundbreaking artificial intelligence (AI) algorithm called FaceAge that can predict a person’s biological age and potentially even their lifespan based on a simple facial photograph. This innovative tool utilizes deep learning to analyze facial features and estimate a subject’s biological age, which reflects the rate at which their body is aging, as opposed to their chronological age, or simply the number of years they’ve lived.

The implications of this technology are far-reaching, particularly in the realm of healthcare. According to a press release from MGB, FaceAge has demonstrated the ability to predict survival outcomes for individuals diagnosed with cancer, suggesting that the tool could offer valuable insights into patient prognosis and treatment planning.

How FaceAge Works

The AI algorithm was trained using an extensive dataset of 58,851 photographs of presumed healthy individuals sourced from publicly available datasets. This training process enabled FaceAge to learn the complex relationships between facial features and age, allowing it to accurately estimate biological age from new images.

To assess FaceAge’s accuracy, researchers applied it to analyze photographs of 6,196 cancer patients taken prior to their radiotherapy treatment. The results revealed that the tool consistently generated a higher biological age for these patients, averaging about five years older than their actual chronological age. This suggests that FaceAge can identify signs of accelerated aging in individuals with cancer, potentially reflecting the impact of the disease on their overall health.

Predicting Life Expectancy

In another experiment, researchers evaluated FaceAge’s ability to predict the life expectancy of 100 patients receiving palliative care. They compared the tool’s predictions with those made by 10 experienced clinicians. Remarkably, FaceAge outperformed the clinicians in accurately predicting patient survival, indicating its potential to provide more objective and precise assessments of life expectancy.

The findings of this research have been published in the esteemed journal The Lancet Digital Health.

Expert Insights

Hugo Aerts, PhD, co-senior and corresponding author of the study, and director of the Artificial Intelligence in Medicine (AIM) program at Mass General Brigham, expressed his excitement about the potential of FaceAge, stating, "We can use artificial intelligence to estimate a person’s biological age from face pictures, and our study shows that information can be clinically meaningful."

He further emphasized that "this work demonstrates that a photo like a simple selfie contains important information that could help to inform clinical decision-making and care plans for patients and clinicians."

Ray Mak, MD, co-senior author and faculty member in the AIM program at Mass General Brigham, echoed Aerts’ enthusiasm, noting, "This opens the door to a whole new realm of biomarker discovery from photographs, and its potential goes far beyond cancer care or predicting age. As we increasingly think of different chronic diseases as diseases of aging, it becomes even more important to be able to accurately predict an individual’s aging trajectory."

Mak added, "I hope we can ultimately use this technology as an early detection system in a variety of applications, within a strong regulatory and ethical framework, to help save lives."

Addressing Bias and Ethical Concerns

While the potential benefits of FaceAge are substantial, researchers acknowledge the importance of addressing potential biases and ethical concerns. Dr. Harvey Castro, a board-certified emergency medicine physician and AI futurist based in Dallas, Texas, shared his insights on the tool, emphasizing both its promise and potential pitfalls.

"As an emergency physician and AI futurist, I see both the promise and peril of AI tools like FaceAge," Castro stated. "What excites me is that FaceAge structures the clinical instinct we call the ‘eyeball test’ — a gut sense of how sick someone looks. Now, machine learning can quantify that assessment with surprising accuracy."

Castro predicts that FaceAge could assist doctors in tailoring treatment plans and prioritizing palliative care in oncology, where resilience is more critical than a patient’s birthdate.

However, Castro cautioned that AI models are only as reliable as the data they are trained on. "If the training data lacks diversity, we risk producing biased results," he warned. He also emphasized that FaceAge should augment human judgment, not replace it.

In addition, Castro raised important ethical questions about the ownership and storage of facial data, as well as the need for patient understanding and consent regarding the analysis of their images. He also noted the potential psychological impact of being told one "looks older" than their age, which could influence treatment decisions and self-perception.

Future Research and Clinical Applications

The researchers at MGB acknowledge that further research is necessary before FaceAge can be widely implemented in clinical settings. Future studies will involve diverse patient populations from multiple hospitals and will focus on cancer patients at various stages of the disease. The research team also plans to investigate FaceAge’s ability to predict other diseases, assess general health status, and estimate lifespan.

The ultimate goal is to develop a robust and reliable tool that can assist clinicians in making more informed decisions, improving patient outcomes, and promoting proactive healthcare.