AD i.d.e.a. – Alzheimer’s Disease integrated dynamic electronic assessment of Cognition
Sebastian Ueckert (1), Elodie L. Plan (1), Kaori Ito (2), Mats O. Karlsson (1), Brian Corrigan (2), Andrew C. Hooker (1)
(1) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden; (2) Global Clinical Pharmacology, Pfizer Inc, Groton, CT, USA
Objectives: Assessing a persons cognitive ability is a challenging and time consuming process, yet essential for the diagnosis and monitoring of patients with Alzheimer’s disease (AD). Existing cognitive tests are either quick, e.g., mini-mental state examination (MMSE) [1], or precise, e.g., ADAS-cog [2], but fail to be both. The objective of this project was to develop a procedure that achieves both, by combining pharmacometric methods with the capabilities of a modern Web application, and creating an integrated dynamic electronic assessment of cognition in Alzheimer’s disease.
Methods:
IRT model: The work was based on an item response theory (IRT) model, which links the response of each test in the ADAS-cog and MMSE assessments to the hidden variable cognitive disability using a binary, binomial, or ordered categorical probability model [3].
Adaptive Test Algorithm: The procedure consists of 3-step iterations: I) an approximate posterior is determined from previous patient responses, the population prior for disability, and the IRT model, II) the expected Fisher information for all cognitive tests in the database is calculated using the posterior, III) the most informative test is selected, presented to the patient, whose response is recorded to obtain refined estimates in the next iteration.
Web Application: Adaptive test selection and storing of patient responses are handled on the server-side through a Ruby on Rails based web application with connection to a SQLite database. On the client-side, the user interface is implemented using HTML5 and JavaScript in a responsive design paradigm to optimize usability for wide a range of devices (from smartphone to desktop).
Results: The AD i.d.e.a. application was created and the adaptive testing algorithm was compared to the MMSE assessment via simulations (n=1000). In all 3 simulated patient populations (mild cognitively impaired, mild AD, and AD) the adaptive algorithm had a lower root mean squared error (Δ=-15.7% on average) and was 33% shorter than the MMSE.
Conclusions: The adaptive algorithm used in the AD i.d.e.a. application improves cognitive testing in AD patients by making it quicker and more precise than existing tests. This project exemplifies how pharmacometric methods can be combined with modern Web technology to be integrated into the clinic and directly affect patient care.
Acknowledgement: This work was supported by the DDMoRe (www.ddmore.eu) project.
References:
[1] Folstein MF, Folstein SE, McHugh PR. "Mini-mental state". A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975 Nov;12(3):189-98.
[2] Rosen WG, Mohs RC, Davis KL. A new rating scale for Alzheimer’s disease. Am J Psychiatry. 1984 Nov;141(11):135664.
[3] Ueckert S, Plan EL, Ito K, Karlsson M, Corrigan B, Hooker AC. Application of Item Response Theory to ADAS-cog Scores Modelling in Alzheimer’s Disease. PAGE 21 (2012) Abstr 2318 [www.page-meeting.org/?abstract=2318]
