Dataset Information

Machine learning for identification of frailty in Canadian primary care practices.

ABSTRACT:

Introduction

Frailty is a medical syndrome, commonly affecting people aged 65 years and over and is characterized by a greater risk of adverse outcomes following illness or injury. Electronic medical records contain a large amount of longitudinal data that can be used for primary care research. Machine learning can fully utilize this wide breadth of data for the detection of diseases and syndromes. The creation of a frailty case definition using machine learning may facilitate early intervention, inform advanced screening tests, and allow for surveillance.

Objectives

The objective of this study was to develop a validated case definition of frailty for the primary care context, using machine learning.

Methods

Physicians participating in the Canadian Primary Care Sentinel Surveillance Network across Canada were asked to retrospectively identify the level of frailty present in a sample of their own patients (total n = 5,466), collected from 2015-2019. Frailty levels were dichotomized using a cut-off of 5. Extracted features included previously prescribed medications, billing codes, and other routinely collected primary care data. We used eight supervised machine learning algorithms, with performance assessed using a hold-out test set. A balanced training dataset was also created by oversampling. Sensitivity analyses considered two alternative dichotomization cut-offs. Model performance was evaluated using area under the receiver-operating characteristic curve, F1, accuracy, sensitivity, specificity, negative predictive value and positive predictive value.

Results

The prevalence of frailty within our sample was 18.4%. Of the eight models developed to identify frail patients, an XGBoost model achieved the highest sensitivity (78.14%) and specificity (74.41%). The balanced training dataset did not improve classification performance. Sensitivity analyses did not show improved performance for cut-offs other than 5.

Conclusion

Supervised machine learning was able to create well performing classification models for frailty. Future research is needed to assess frailty inter-rater reliability, and link multiple data sources for frailty identification.

SUBMITTER: Aponte-Hao S

PROVIDER: S-EPMC8431345 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:The prompt identification of frailty in primary care is the first step to offer personalized care to older individuals. We aimed to detect and quantify frailty among primary care older patients, by developing and validating a primary care frailty index (PC-FI) based on routinely collected health records and providing sex-specific frailty charts. The PC-FI was developed using data from 308,280 primary care patients ≥ 60 years old part of the Health Search Database (HSD) in Italy (baseline 2013-2019) and validated in the Swedish National Study on Aging and Care in Kungsholmen (SNAC-K; baseline 2001-2004), a well-characterized population-based cohort including 3363 individuals ≥ 60 years old. Potential health deficits part of the PC-FI were identified through ICD-9, ATC, and exemption codes and selected through an optimization algorithm (i.e., genetic algorithm), using all-cause mortality as the main outcome for the PC-FI development. The PC-FI association at 1, 3 and 5 years, and discriminative ability for mortality and hospitalization were tested in Cox models. The convergent validity with frailty-related measures was verified in SNAC-K. The following cut-offs were used to define absent, mild, moderate and severe frailty: < 0.07, 0.07-0.14, 0.14-0.21, and ≥ 0.21. Mean age of HSD and SNAC-K participants was 71.0 years (55.4% females). The PC-FI included 25 health deficits and showed an independent association with mortality (hazard ratio range 2.03-2.27; p < 0.05) and hospitalization (hazard ratio range 1.25-1.64; p < 0.05) and a fair-to-good discriminative ability (c-statistics range 0.74-0.84 for mortality and 0.59-0.69 for hospitalization). In HSD 34.2%, 10.9% and 3.8% were deemed mildly, moderately, and severely frail, respectively. In the SNAC-K cohort, the associations between PC-FI and mortality and hospitalization were stronger than in the HSD and PC-FI scores were associated with physical frailty (odds ratio 4.25 for each 0.1 increase; p < 0.05; area under the curve 0.84), poor physical performance, disability, injurious falls, and dementia. Almost 15% of primary care patients ≥ 60 years old are affected by moderate or severe frailty in Italy. We propose a reliable, automated, and easily implementable frailty index that can be used to screen the primary care population for frailty.