AI Tool Predicts Aging & Health in Cancer Patients from Facial Features

Researchers have developed a new AI tool called FaceAge, which relies on analyzing facial images to estimate biological age and predict survival rates in cancer patients. In a special study to test the tool, conducted on more than 6,000 patients, the results showed that the estimated biological age of these patients was, on average, about five years older than their actual ages, and this difference was associated with worse health outcomes.

The tool was distinguished by its accuracy, surpassing doctors in predicting the average short-term survival, especially in patients undergoing palliative radiation therapy, which enhances its role as a supportive tool in making medical decisions. The study results indicate that facial features may be accurate vital indicators for predicting aging and the health status of patients, paving the way for wide applications in precision medicine (personalized medicine).

Study Details A team of researchers at Mass General Brigham developed a deep learning algorithm known as FaceAge, which relies on facial images to predict the biological age and health outcomes of cancer patients.

The results showed that the patients who participated in the study, numbering 6,196 individuals, appeared older in their images than they actually were, as their estimated biological age exceeded their chronological age by about five years. It was also observed that patients for whom the tool estimated higher biological ages had lower chances of survival, especially in cases where the biological estimate exceeded the 85-year mark.

In another experiment, ten doctors were asked to assess the average survival of 100 patients receiving palliative treatment, based on their images and medical data. Their predictions were less accurate than the predictions of the FaceAge AI tool, but when they were provided with the tool’s estimates of biological ages, their predictions improved significantly.

Promising Technology and Future Applications The team relied on deep learning and facial analysis techniques in developing the FaceAge AI tool, and they trained the algorithm using more than 58,000 images of supposedly healthy individuals from public databases. They then tested the algorithm on a group of 6,196 cancer patients in two medical centers, using images taken at the start of radiation therapy.

The study results were published in The Lancet Digital Health journal, indicating great potential for this technology in modern medicine. Dr. Hugo Aerts, one of the study’s authors, says, “A simple image can reveal a lot about a person’s biological health, and our study has shown that the consistency between a person’s appearance and their actual age carries important clinical implications.”

He added that people who appear younger than their age achieve better results after cancer treatment, reflecting the importance of visual indicators in health assessment.

For his part, Dr. Ray Mak, one of the study participants, pointed out that this innovation opens the field for a new era of discovering vital indicators from photographs, and its applications go beyond estimating age or treating cancer to include many chronic diseases associated with aging.

Ethical and Religious Considerations Even with the high predictive accuracy of the FaceAge tool for patients’ biological age, it is necessary to know that predicting age or time of death is not a certainty, but rather remains within the scope of probabilistic estimates based on data, as ages are in the hands of God alone and the date of death remains a matter of the unseen. Therefore, this tool should be treated as an aid in supporting healthcare decisions and used as an early warning system in various applications, within a strict regulatory and ethical framework.