Feasibility study for the use of computer game base AI technology in the early diagnosis of Mild Cognitive Impairment Alzheimer’s Disease

There is a chronic lack of an inexpensive, simple and effective way to accurately screen and diagnose Alzheimer s Disease: Affecting 55Mn people globally[1], rising to over 150Mn by 2050[3]. 41Mn people remain undiagnosed, hence no access to appropriate care[2].

In the UK at least 20% are diagnosed inaccurately[5] Costing the global healthcare system $1Tn annually, tripling by 2030[3;7] £34.7Bn UK alone, increasing to £94.1Bn by 2040[4].

Indicating massive and growing unmet need[1;6;7] to identify individuals easily and early to allocate management/care-resources accordingly.

Current gold-standard diagnostics are reliant on initial GP-screening then consequent multiple protracted clinical diagnostic examinations by numerous healthcare professionals.

These methods are highly resource intensive and expensive in terms of clinical time, effort, burden and complexity (need for multiple tests).

For the patients these tests are also highly invasive (Spinal puncture, blood, PET and MRI scans), obviously daunting and suffer from a substantially growing waiting-list between initial GP-screening and eventual hospital diagnosis. Furthermore, accessibility; 52% fear diagnosis, 56% delay seeking diagnosis for up-to a year or more[8;9].

Current State-Of-The-Art[10;11;12;13] demand unfeasibly high user-motivation, attention, are symptom-based, lack evidence, require costly specialist hardware, installation, training, maintenance and multiple operators, proving ineffective.

We have developed a highly-innovative deep-learning-Artificial-Intelligence(AiaMD)-game-based proof-of-concept-TRL3-prototype currently being clinically validated.

Targeting collaborator newly-discovered braincells that are the earliest affected by Alzheimers-Disease[14-High-Profile-Nature-Neuroscience] providing substantially earlier and much more accurate concurrent primary-care-GP-screening and secondary-care-diagnostics over anything currently available, potentially leading towards a world-first non-pharmacological treatment.

Current testing demonstrates substantially-higher efficacy (sensitivity and specificity) between healthy, early-Alzheimer s and mild-cognitively-impaired patients versus RAVLT, ACE3, MMSE and MoCA gold-standard clinical-tests in 3 separate gameplay-tests featuring 20 PPI and 18 clinical volunteers, informing test design/outcome measures and proving usability/effectiveness.

This project will confirm previous PPI software modifications and guide further development from our £300k IUK-Smart project ensuring optimal usability/end-user correctness of our videogame-prototype.