A Machine Learning–Based Predictive Model to Identify Patients Who Failed to Attend a Follow-up Visit for Diabetes Care After Recommendations From a National Screening Program

Abstract

OBJECTIVE Reportedly, two-thirds of the patients who were positive for diabetes during screening failed to attend a follow-up visit for diabetes care in Japan. We aimed to develop a machine-learning model for predicting people’s failure to attend a follow-up visit. RESEARCH DESIGN AND METHODS We conducted a retrospective cohort study of adults with newly screened diabetes at a national screening program using a large Japanese insurance claims database (JMDC, Tokyo, Japan). We defined failure to attend a follow-up visit for diabetes care as no physician consultation during the 6 months after the screening. The candidate predictors were patient demographics, comorbidities, and medication history. In the training set (randomly selected 80% of the sample), we developed two models (previously reported logistic regression model and Lasso regression model). In the test set (remaining 20%), prediction performance was examined. RESULTS We identified 10,645 patients, including 5,450 patients who failed to attend follow-up visits for diabetes care. The Lasso regression model using four predictors had a better discrimination ability than the previously reported logistic regression model using 13 predictors (C-statistic: 0.71 [95% CI 0.69–0.73] vs. 0.67 [0.65–0.69]; P < 0.001). The four selected predictors in the Lasso regression model were lower frequency of physician visits in the previous year, lower HbA1c levels, and negative history of antidyslipidemic or antihypertensive treatment. CONCLUSIONS The developed machine-learning model using four predictors had a good predictive ability to identify patients who failed to attend a follow-up visit for diabetes care after a screening program.

Impact of the COVID-19 Pandemic on Health Check-ups in 2021 and 2022: A Nationwide Follow-up Survey of Healthcare Facilities in Japan Society of Ningen Dock

“…Adherence to follow-up visits among those who were required to have them varied depending on the type of screening. Adherence was higher for cancer screenings than for diabetes, hypertension, or dyslipidemia screenings, which is consistent with previous reports (11) , (14) , (15) , (16) . Adherence improved for some cancer screenings, as well as for diabetes, hypertension, and dyslipidemia screenings in 2021, suggesting that those who underwent screenings in 2021 were more likely to have higher health literacy.…”

Proportion of subsequent clinic visits among persons without regular clinic visits who were screened as having hyperglycemia: A retrospective cohort study

“…The positive association between the HbA1c levels at the health checkup and the clinic visits in the present study was also reported by Tsujimura et al 10 . and Okada et al ., 13 whereas the influence of HbA1c level on the subsequent clinic visit rates hit a low value ceiling and was <30% even for HbA1c ≥8.0%. We added further evidence that previous health checkup results can affect the subsequent clinic visit.…”

“…Tsujimura et al included glucose measurement in their definition of clinic visit, and we did not. Recently, Okada et al 13 showed that fewer physician visits in the previous year, lower HbA1c levels and no history of antidyslipidemic or antihypertensive treatment were associated with fewer clinic visits. They included the presence/absence of hyperglycemia in the health checkups a year before as a variable in their candidate predictors; however, the Lasso regression estimating the proportion of follow-up clinic visits did not choose the variable in the final model.…”

“…included glucose measurement in their definition of clinic visit, and we did not. Recently, Okada et al 13 . showed that fewer physician visits in the previous year, lower HbA1c levels and no history of antidyslipidemic or antihypertensive treatment were associated with fewer clinic visits.…”

“…Because early intervention for suspected type 2 diabetes is effective in inhibiting disease progression 11 , and a previous study showed a decrease in HbA1c levels among those who had follow‐up clinic visits after health checkups 12 , improving the approach to recommending clinical visits is imperative. In a recent study, the frequency of clinic visits in the year before the health checkup was associated with the likelihood of follow‐up clinic visits after health checkups 13 . In this analysis, the patient sample was heterogeneous in terms of access to medical services; thus, it is quite possible that those who had been followed up by a physician for other chronic diseases received diabetes care after it was detected during health checkups.…”