Researchers in the United States have developed an artificial intelligence system that uses changes in facial appearance over time to help predict how cancer patients may respond to treatment and their chances of survival.

The tool, known as FaceAge, was created by scientists at Mass General Brigham. It estimates a person’s biological age from a photograph, offering a measure of how old the body appears physiologically rather than simply counting years since birth.

Biological age can differ from chronological age, as factors such as illness, stress and lifestyle often accelerate the ageing process. Researchers say facial features may provide important clues about a person’s overall health.

Earlier studies using FaceAge found that cancer patients typically appeared about five years older biologically than their actual age. Those with older-looking facial profiles were also more likely to experience poorer outcomes following treatment.

In the latest research, scientists introduced a new metric called Face Aging Rate, or FAR, which tracks changes in biological age over time by comparing multiple photographs. The method is designed to monitor how quickly a person appears to age, potentially offering a real-time indicator of health.

The study analysed images of 2,276 cancer patients treated at Brigham and Women’s Hospital between 2012 and 2023. All participants had undergone at least two courses of radiation therapy, with photographs taken routinely during their treatment.

Researchers found that, on average, patients’ facial ageing progressed about 40% faster than their actual chronological ageing. Those with higher FAR scores had significantly lower survival rates, particularly when the photographs were taken more than two years apart.

The findings suggest that changes in facial appearance over time may provide valuable insight into a patient’s condition and long-term prognosis.

The study also examined FaceAge Deviation, a separate measure comparing biological age from a single photograph with a person’s actual age. While higher scores were also linked to poorer outcomes, FAR proved to be a stronger and more reliable predictor of survival over extended periods.

Researchers believe combining both measures could offer a more complete picture of a patient’s health and disease progression.

Dr Raymond Mak, a radiation oncologist at Mass General Brigham Cancer Institute, said the technology could help doctors refine treatment plans, improve patient counselling and determine the most appropriate follow-up care.

The team also sees broader potential beyond oncology. Future research will explore whether the technology could help assess other chronic illnesses or even provide early health insights for otherwise healthy individuals.

To support ongoing studies, researchers have launched a public web portal where users can upload a photograph, receive a FaceAge estimate and contribute data to further development of the tool.

Artificial intelligence could soon help doctors identify children at risk of attention-deficit/hyperactivity disorder (ADHD) years before they receive a formal diagnosis, according to new research that could transform early intervention for one of childhood’s most common neurodevelopmental disorders.

The study, conducted by researchers at Duke Health and published in Nature Mental Health, found that AI can analyse routine electronic health records to detect early warning signs of ADHD long before symptoms typically lead to a diagnosis.

ADHD affects an estimated 8% of children and adolescents. It is characterised by symptoms such as difficulty concentrating, impulsivity, restlessness and problems with organisation. Despite its prevalence, many children are not diagnosed until years after symptoms first appear, delaying access to treatment and support.

Researchers examined health records from more than 140,000 children, including both those diagnosed with ADHD and those without the condition. Using data collected from birth through early childhood, the AI model was trained to identify patterns linked to later ADHD diagnoses.

The system successfully recognised combinations of developmental, behavioural and clinical indicators that often emerged years before a formal diagnosis. It was particularly accurate in assessing risk among children aged five and older, and its performance remained consistent across sex, race, ethnicity and insurance status.

Elliot Hill, the study’s lead author and a data scientist at Duke University School of Medicine, said electronic health records contain a wealth of information that can reveal important patterns.

The researchers believe the technology could help clinicians identify children who may benefit from earlier assessment, allowing families to access support sooner. Early intervention has been shown to improve academic performance, social development and long-term health outcomes.

Naomi Davis, an associate professor in Duke’s Department of Psychiatry and Behavioral Sciences and a co-author of the study, said timely support is critical for children with ADHD.

“Children with ADHD can really struggle when their needs aren’t understood and adequate supports are not in place,” she said.

The research team stressed that the tool is not intended to replace doctors or provide a definitive diagnosis. Instead, it is designed to assist clinicians by highlighting children who may require further evaluation.

Matthew Engelhard, senior author of the study, described it as a way to ensure that children who need help are identified earlier and do not face unnecessary delays.

Experts say the approach could eventually be expanded to other areas of mental health, with similar AI models already being explored to better understand psychiatric risks in adolescents.

The findings also highlight the potential to improve recognition of ADHD in girls, who are often underdiagnosed because their symptoms tend to be less overt than those seen in boys.