Non-Invasive Solutions to Identify Distinctions between Healthy and Mild Cognitive Impairments Participants

Abstract

Mild cognitive impairment (MCI) is a condition characterized by impairment in a single cognitive domain or mild deficit in several cognitive domains. MCI patients are at increased risk of progression to dementia with almost 50% of MCI patients developing dementia within five years. Early detection can play an important role in early intervention, prevention, and appropriate treatments. In this study, we examined heart rate variability (HRV) as a novel physiological biomarker for identifying individuals at higher risk of MCI. We investigated if measuring HRV using non-invasive sensors might offer reliable, non-invasive techniques to distinguish MCI patients from healthy controls. Twenty-one MCI patients were recruited to examine this possibility. HRV was assessed using CorSense wearable device. HRV indices were analyzed and compared in rest between MCI and healthy controls. The significance of difference of numerical data between two groups was assessed using parametric unpaired t-test or non-parametric Wilcoxon rank sum test based on the fulfilment of unpaired t-test assumptions. Multiple linear regression models were performed to assess the association between individual HRV parameter with the cognitive status adjusting for gender and age. Time-domain parameters i.e., the standard deviation of NN intervals (SDNN), and the root mean square of successive differences between normal heartbeats (RMSSD) were significantly lower in MCI patients compared with healthy controls. Prediction accuracy for the logistic regression using 10-fold cross-validation was 76.5%, Specificity was 0.8571, while sensitivity was 0.8095. Our study demonstrated that healthy participants have higher HRV indices compared to older adults with MCI using non-invasive biosensors technologies. Our results are of clinical importance in terms of showing the possibility that MCI of older people can be predicted using only HRV PPG-based data.