Researchers develop AI and machine learning algorithms to streamline dementia diagnoses
Researchers have redesigned decades-old dementia diagnostic assessments to streamline the diagnosis process for patients. Statistics from The World Health Organization describe dementia as the seventh leading cause of death among all diseases and a major cause of disability and dependency among older people, despite this, there is currently an underdiagnosis and misdiagnosis of the disease.
An interdisciplinary team of researchers from Ulster University, WHSCT, Northern Ireland Centre for Stratified Medicine and NUI Galway have developed AI and machine learning algorithms to leads to create a dementia diagnosis process that is not only predictive but also efficient in terms of overall administration time. To facilitate engagement between patients, clinicians and policy makers, a sandbox interface app has been developed to enhance the diagnosis process, with a video demo available online.
This original research work has been published in the IEEE Journal of Translational Engineering in Health and Medicine. The lead researcher and author from the Intelligent Systems Research Centre at Ulster University, Dr. Niamh McCombe, commented:
Underdiagnosis and misdiagnosis of dementia is devastating for patients and their families, and we sought to cutting edge technology to help solve this problem. We understand that doctors in primary care have only limited consultation time with patients and by introducing AI and machine learning algorithms to streamline the diagnosis process and ensure support and treatment provision is swift.
An extension of this work by considering different types of costs such as financial costs of assessments will be presented as a conference paper at an upcoming conference.
Senior author of these studies, Dr. KongFatt Wong-Lin, added:
This work aligns with the current trend in AI and machine learning that moves towards more practical use by a wider range of people, a so-called citizen science. For our work, our app helps tap into the domain expertise of clinicians. Clinicians can, for instance, apply their extensive knowledge of their field and local clinical practice to customise the algorithms in the app for their own use. Such ‘humans-in-the-loop’ algorithms can lead to greater trustworthy and adoption from non-AI specialist users.”
Dr. Joseph Kane, a clinical academic lecturer at Queen’s University Belfast and Honorary Consultant in Psychiatry of Old Age at Belfast Health and Social Care Trust, who was not involved in the research studies, remarked:
The fields of AI and machine learning are providing promising new insights into the detection of dementia. These impressive work by McCombe and colleagues has used such methods to explore what many consider to be a critical obstacle to cognitive assessment in primary care – the time and financial cost associated with administering diagnostic tools and tests. The work has highlighted the importance in engaging clinicians and other stakeholders in translating AI from the laboratory to the clinical room.”