Event Title

Correlation between CSF biomarkers of Alzheimer's disease and cognitive decline toward a machine learning based predictive model

Presenter Information

Vivek Kumar TiwariFollow

Loading...

Media is loading
 

Faculty Mentor

Dr.Shawana Tabassum & Dr. Premananda Indic

Document Type

Oral Presentation

Date of Publication

4-16-2021

Abstract

Early diagnosis of Alzheimer disease (AD) is still lacking as the traditional cognitive tests have several limitations. The most commonly utilized Mini–Mental State Examination (MMSE) scores assess cognition at one point in time and do not reflect its decline over time, they do not assess the subject's functional status, and are susceptible to cultural influences. The search for molecular biomarkers for precise and early detection of the AD stages remains a challenge. The three biomarkers namely Ab1-42, T-tau, and P-tau that are found in the cerebrospinal fluid (CSF), have shown promise in AD diagnosis. In this work, we have analyzed an electronic health record of 378 subjects (collected from the National Alzheimer's Coordinating Center database), including 145 subjects with normal cognition, 105 with mild dementia, 104 with moderate dementia, and 24 with severe dementia. We calculated the association between CSF biomarkers of the subjects with their MMSE scores using Pearson correlation. Our results : 1) in subjects with moderate dementia, MMSE scores correlate weakly with the three CSF biomarkers (r=0.19 for Ab1-42, r=0.15 for P-tau, and r=0.13 for T-tau) and 2) in subjects with severe dementia MMSE correlate moderately with the biomarkers (r=-0.34 for Ab1-42 and r= -0.62 for T-tau). The results are quite promising as they validate the need for a point-of-care sensor for non-invasive monitoring of these biomarker levels over time to facilitate early diagnosis and treatment of AD. Our next goal is to train a ML model for predicting the stage and conducting risk stratification of AD from the biomarker levels. The impact of our research is significant because the prediction models will aid the clinicians in diagnosing AD early and taking preventive action accordingly.

Keywords

Alzheimer disease, cerebrospinal fluid biomarkers, Machine learning model

Persistent Identifier

http://hdl.handle.net/10950/3132

This document is currently not available here.

Share

COinS
 
Apr 16th, 9:00 AM Apr 16th, 10:00 AM

Correlation between CSF biomarkers of Alzheimer's disease and cognitive decline toward a machine learning based predictive model

Early diagnosis of Alzheimer disease (AD) is still lacking as the traditional cognitive tests have several limitations. The most commonly utilized Mini–Mental State Examination (MMSE) scores assess cognition at one point in time and do not reflect its decline over time, they do not assess the subject's functional status, and are susceptible to cultural influences. The search for molecular biomarkers for precise and early detection of the AD stages remains a challenge. The three biomarkers namely Ab1-42, T-tau, and P-tau that are found in the cerebrospinal fluid (CSF), have shown promise in AD diagnosis. In this work, we have analyzed an electronic health record of 378 subjects (collected from the National Alzheimer's Coordinating Center database), including 145 subjects with normal cognition, 105 with mild dementia, 104 with moderate dementia, and 24 with severe dementia. We calculated the association between CSF biomarkers of the subjects with their MMSE scores using Pearson correlation. Our results : 1) in subjects with moderate dementia, MMSE scores correlate weakly with the three CSF biomarkers (r=0.19 for Ab1-42, r=0.15 for P-tau, and r=0.13 for T-tau) and 2) in subjects with severe dementia MMSE correlate moderately with the biomarkers (r=-0.34 for Ab1-42 and r= -0.62 for T-tau). The results are quite promising as they validate the need for a point-of-care sensor for non-invasive monitoring of these biomarker levels over time to facilitate early diagnosis and treatment of AD. Our next goal is to train a ML model for predicting the stage and conducting risk stratification of AD from the biomarker levels. The impact of our research is significant because the prediction models will aid the clinicians in diagnosing AD early and taking preventive action accordingly.