Public Health.

Area of Science:

• Neurology
• Artificial Intelligence
• Digital Health

Background:

• Early detection of mild neurocognitive disorder (NCD) is crucial for timely interventions.
• Progression to major neurocognitive disorder can be prevented with early diagnosis.
• This study focuses on developing a digital screening tool for mild NCD detection.

Purpose of the Study:

• To develop a sensitive and simple digitalized screening tool for mild NCD.
• To utilize artificial intelligence for capturing and interpreting spoken language.
• To enable early detection of mild NCD before progression.

Main Methods:

• Developed the Digitalised Combined Objective-Subjective (DCOS) cognitive screening tool based on DSM-5 criteria.
• Validated the tool on two independent cohorts (n=977 and n=127).
• Selected DCOS items using ROC analysis from existing cognitive assessments, focusing on spoken language domains.

Main Results:

• The DCOS tool comprises six items (three subjective, three objective) with high diagnostic accuracy (AUC=0.87 in validation).
• DCOS demonstrated superior performance compared to Ascertain Dementia 8 (AD8) and comparable accuracy to HK MoCA-5min.
• The tool achieved 88% sensitivity and 69% specificity, showing feasibility for digitalization.

Conclusions:

• DCOS is a novel screening tool assessing both subjective and objective cognitive domains for mild NCD.
• Its brief administration time (<5 minutes) makes it suitable for primary healthcare settings.
• Future integration with AI on a digital platform will enhance large-scale utility.