Approved research

Optimising Blood Biomarker Accuracy for Clinical Trials and Healthcare using Artificial Intelligence and Digital Biomarkers

Lay summary

The need for accurate early diagnosis

The 2022 World Alzheimer Report estimates that 55 million people globally are living with dementia, projected to rise to 139 million by 2050. Early diagnosis tools are desperately needed, with diagnosis typically occurring late in the disease process and misdiagnosis common due to the overlap of cognitive symptoms in the early stages of dementia. Earlier, more accurate diagnosis would allow delivery of both symptom and disease modifying treatments at a more effective stage, and aid drug development through more accurate stratification of patients in clinical trials.

Existing biomarkers

Biomarkers of dementia are available, but many are costly and difficult to access or require painful medical procedures. Blood biomarkers are a recent development that provide a great opportunity for cheap, easy-to-access biomarkers that most people would be comfortable providing. However, while blood biomarkers are an important part of the dementia biomarker future, they are not perfect. We are learning through research that their accuracy can vary depending on how well you sleep, your ethnicity, even what time of day the samples were taken.

Improving blood biomarker accuracy

Cumulus Neuroscience is a UK company, and a global leader in dementia digital biomarker development. This project will help to improve the accuracy of blood biomarkers by developing an Artificial Intelligence (AI) diagnostic tool, that combines blood biomarkers with digital biomarkers (direct measures of brain function developed by Cumulus) and demographic information. By combining different data types we will get a more complete picture of an individual's brain health and learn which are the most important features on which to base a classification decision. We will then test and validate this algorithm on two real-world data sets from dementia diagnosis research studies Cumulus are conducting. Real-world clinical data is often imperfect and incomplete. Our algorithms will be developed from the outset to be able to make accurate classification decisions even when faced with such imperfect, real-world data. The algorithms will also convey how confident their decision is, helping clinicians to use them most effectively.

Who benefits?

Overall, this project will take the most promising dementia biomarkers and produce a sensitive and accurate AI diagnostic tool that will benefit patients, clinicians and companies developing new treatments. This will allow clinicians to make better decisions about a patient's care, and help industry choose the right patients for drug trials, leading to more effective therapies being approved for use more quickly.