Affiliations: [a] Department of Industrial and Management Systems Engineering, University of South Florida, Tampa, FL, USA | [b] Byrd Alzheimer's Institute, University of South Florida, Tampa, FL, USA | [c] School of Aging Studies, University of South Florida, Tampa, FL, USA
Correspondence:
[*]
Correspondence to: Shuai Huang, PhD, Department of Industrial and Management Systems Engineering, University of South Florida, 4202 E Fowler Ave, ENC 2509, Tampa, FL 33620, USA. Tel.: +1 813 974 2090; Fax: +1 813 974 5953; E-mail: [email protected].
Note: [1] Data used in preparation of this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (http://adni.loni.usc.edu/). As such, the investigators within the ADNI contributed to the design and implementation of ADNI and/or provided data but did not participate in analysis or writing of this report. A complete listing of ADNI investigators can be found at: http://adni.loni.usc.edu/wp-content/uploads/how_to_apply/ADNI_Acknowledgement_List.pdf
Abstract: Background:Detecting participants who are positive for amyloid-β (Aβ) pathology is germane in designing prevention trials by enriching for those cases that are more likely to be amyloid positive. Existing brain amyloid measurement techniques, such as the Pittsburgh Compound B-positron emission tomography and cerebrospinal fluid, are not reasonable first-line approaches limited by either feasibility or cost. Objective:We aimed to identify simple and cost-effective rules that can predict brain Aβ level by integrating both neuropsychological measurements and blood-based markers. Method:Several decision tree models were built for extracting the predictive rules based on the Alzheimer's Disease Neuroimaging Initiative cohort. Results:We successfully extracted predictive rules of Aβ level. For cognitive function variables, cases above the 45th percentile in total cognitive score (TOTALMOD), above the 52nd percentile of delayed word recall, and above the 70th percentile in orientation resulted in a group that was highly enriched for amyloid negative cases. Conversely scoring below the 15th percentile of TOTALMOD resulted in a group highly enriched for amyloid positive cases. For blood protein markers, scoring below the 57th percentile for apolipoprotein E (ApoE) levels (irrespective of genotype) enriched two fold for the risk of being amyloid positive. In the high ApoE cases, scoring above the 60th percentile for transthyretin resulted in a group that was >90% amyloid negative. A third decision tree using both cognitive and blood-marker data slightly improved the classification of cases. Conclusion:Our study demonstrated that the integration of the neuropsychological measurements and blood-based markers significantly improved prediction accuracy. The prediction model has led to several simple rules, which have a great potential of being naturally translated into clinical settings such as enrichment screening for AD prevention trials of anti-amyloid treatments.
