Can AI Predict Disease Before Symptoms Appear?

Artificial Intelligence has advanced to the point where it can recognise patterns that are not visible to the human eye. One example is the analysis of blood samples through machine learning models which have been trained to recognise certain protein signatures with characteristics of certain diseases. 

Can AI Really Predict Illness?

Yes, AI can really predict illness well in advance. One notable example is the work done by the University of Edinburgh, where AI was used to analyse blood samples of tens of thousands of test subjects. Using AI, the researchers were able to remotely predict the onset of Alzheimer’s and other related diseases, assumed to detect the disease 10 years earlier than symptoms would generally appear. 

How Can AI Predict Alzheimer’s Disease?

The AI models developed by the University of Edinburgh were more advanced than other models which were based on common risk factors such as age, lifestyle and cholesterol levels, as they were able to recognise minute protein patterns that are often ignored by common medical tests.

What Other Diseases Can Be Detected with This Blood Sample?

This advanced AI testing isn’t just proving useful for early detection for Alzheimer’s Disease, but in the instance of Parkinson’s disease, scientists have been able to train algorithms to identify the pattern of only eight proteins in a blood sample to predict the disease seven years, or more, prior to symptoms showing. 

How Could AI Help in Predicting and Preventing Diseases Before They Occur?

As AI begins to discover disease early-on, we can begin taking preventative measures such as closer monitoring, preventative lifestyle changes and individualised treatment plans aimed at delaying or possibly preventing the disease from coming to fruition. AI enhanced screening could be incorporated into standard health checks which allows physicians to identify and mitigate ill-health risk at the pre- or developed disease phase. This changes the practice of medicine to being proactive and not just reactive.

Med-Tech in Action: Blood Tests Through A Pin-Prick

One of the most promising recent developments comes from research that analyses blood samples taken via a small pin prick. These samples are the same as those taken when checking glucose levels on a routine basis for diabetics. These samples are analysed with the help of artificial intelligence in order to metabolically profile and determine the presence of hundreds of metabolic compounds in the samples. One study from a UK Biobank showed that metabolic analyses of blood samples predicted certain patterns linked to diseases like dementia, heart-disease and cancer up to ten years before symptoms would normally appear. 

How Can AI Assist in Predicting and Preventing Disease In the NHS?

Fully embedding AI in NHS systems could mean more timely diagnosis, lower long-term costs of treatment and improved health of the population. With risk markers, GPs could make more informed lifestyle recommendations, create tailored screenings or send patients to preventive interventions before the disease is able to take hold.

For the NHS which is often overextended, being able to predict disease early allows a decrease in demand for complex care. This in turn eases the burden on hospitals, specialist positions and long-term treatment services.

Startups and Funding in the AI Predictive Medicine Space

Investors are becoming engaged with more and more health-tech AI startups in the UK focusing on predicting disease prior to its symptoms, such as Presymptom Health, a spinout developing AI infection and sepsis diagnostic tests for the NHS and recently obtained £1.5 million in seed and grant financing. Their early warning technology is currently being trialed in hospitals within the NHS with plans to expand in the UK and internationally.

Another interesting company is Mendelian, who are also notable for their AI-powered rare disease diagnostics. Their MendelScan solution available in NHS GP practices recently won €1.58 million from NHS AI Lab/NIHR for the Accelerate deployment program, part of the AI in Health and Care Awards.

These are just two examples of how the AI predictive medicine space is gaining traction and significant funding for their works.