Importance of handgrip strength and endurance time for predicting COVID-19 mortality in older adult patients: K-Nearest Neighbors (k-NN) algorithm

Abstract

The study aimed to investigate the role of handgrip strength (HGS) and muscular endurance time (ET), as assessment measures for physical frailty and muscle function, in predicting the COVID-19 mortality of elderly patients admitted to the intensive care unit (ICU). This prospective observational cross-sectional study was conducted on 872 COVID-19 patients (415 females and 457 Males) aged 65-90 years admitted to ICU. Demographic data, underlying comorbidities, COVID-19-related symptoms, as well as laboratory and computed tomography (CT) findings were obtained from the patient's medical records. Using a JAMAR® hydraulic dynamometer, the average HGS (kg) after three measurements on the dominant side was recorded as the outcome for analysis. The threshold of the Low grip strength was defined as less than 26 kg and 14 kg for males and females, respectively. This is based on the consideration that low grip is two standard deviations below the gender-specific peak mean value. Muscular ET was also calculated after an additional trial, in which patients were asked to maintain the grip, and the value was measured in seconds when strength dropped to 50% of its maximum level. Subsequently, all thirty-one features were entered into the k-Nearest Neighbors (k-NN) algorithm to investigate the possible relationship between HGS and ET with COVID-19 mortality in elderly patients admitted to ICU. The results showed that chronic obstructive pulmonary disease (COPD), low grip strength, C-reactive protein (CRP), SaO2, and ET were found to be the most relevant components for possible COVID-19 mortality prediction, respectively. Further, the k-NN classifier achieved the highest classification accuracy of 95.21% to predict COVID-19 mortality, under the 10-fold data division protocol. Along with the well-known clinical risk factors, HGS and ET can be quick and low-cost prognostic tools in the mortality rate of elderly patients with COVID-19.

Keywords: COVID-19 mortality, Older adults, Muscle strength, Machine learning, Muscular endurance time.

DOI: 10.54941/ahfe1004900