Why Not? Persuading Clinicians to Reduce Overuse

Machine learning and deep learning for classifying the justification of brain CT referrals

OBJECTIVES

To train the machine and deep learning models to automate the justification analysis of radiology referrals in accordance with iGuide categorisation, and to determine if prediction models can generalise across multiple clinical sites and outperform human experts.

METHODS

Adult brain computed tomography (CT) referrals from scans performed in three CT centres in Ireland in 2020 and 2021 were retrospectively collected. Two radiographers analysed the justification of 3000 randomly selected referrals using iGuide, with two consultant radiologists analysing the referrals with disagreement. Insufficient or duplicate referrals were discarded. The inter-rater agreement among radiographers and consultants was computed. A random split (4:1) was performed to apply machine learning (ML) and deep learning (DL) techniques to unstructured clinical indications to automate retrospective justification auditing with multi-class classification. The accuracy and macro-averaged F1 score of the best-performing classifier of each type on the training set were computed on the test set.

RESULTS

42 referrals were ignored. 1909 (64.5%) referrals were justified, 811 (27.4%) were potentially justified, and 238 (8.1%) were unjustified. The agreement between radiographers (κ = 0.268) was lower than radiologists (κ = 0.460). The best-performing ML model was the bag-of-words-based gradient-boosting classifier achieving a 94.4% accuracy and a macro F1 of 0.94. DL models were inferior, with bi-directional long short-term memory achieving 92.3% accuracy, a macro F1 of 0.92, and outperforming multilayer perceptrons.

CONCLUSION

Interpreting unstructured clinical indications is challenging necessitating clinical decision support. ML and DL can generalise across multiple clinical sites, outperform human experts, and be used as an artificial intelligence-based iGuide interpreter when retrospectively vetting radiology referrals.

CLINICAL RELEVANCE STATEMENT

Healthcare vendors and clinical sites should consider developing and utilising artificial intelligence-enabled systems for justifying medical exposures. This would enable better implementation of imaging referral guidelines in clinical practices and reduce population dose burden, CT waiting lists, and wasteful use of resources.

KEY POINTS

Significant variations exist among human experts in interpreting unstructured clinical indications/patient presentations. Machine and deep learning can automate the justification analysis of radiology referrals according to iGuide categorisation. Machine and deep learning can improve retrospective and prospective justification auditing for better implementation of imaging referral guidelines.

What Promotes Medical Overuse: Perspective on Evolutionary Game between Administration and Medical Institutions

Medical overuse is the leading cause of high expenditure among healthcare systems worldwide, with the degree varying from region to region. There is increasing evidence to indicate that in China, National Healthcare Security Administration (NHSA) supervision plays the most crucial role in decreasing medical overuse. For medical overuse, traditional studies focus on empirical researches and qualitative analysis, most of which ignore how the two important participants, i.e., medical institutions and NHSA, affect the strategy of each other. To reduce the losses incurred by insufficient supervision, this study starts from bounded rationality, builds an evolutionary game model to study the relations between the NHSA and medical institutions, and reveals the dynamic evolution process of the supervision of NHSA and overuse of medical institutions. Through stable evolutionary strategy analysis, numerical simulation results, and sensitive experiments under diverse scenarios, we found that when profit gap of medical overuse is high or low, medical institution will adopt fixed strategy, which is medical overuse or appropriate medical use. Only when the profit gap is at a medium level will NHSA's choice affects medical institutions' strategy. Furthermore, NHSA's strategy is affected by the profit gap between medical use and supervision cost. Our work enriches the understanding of supervision for medical overuse and provides theoretical support for the NHSA to make decisions to reach an ideal condition, i.e., to supervise without exertion.

Low-value care and excess out-of-pocket expenditure among older adults with incident cancer - A machine learning approach

OBJECTIVE

To evaluate the association of low-value care with excess out-of-pocket expenditure among older adults diagnosed with incident breast, prostate, colorectal cancers, and Non-Hodgkin's Lymphoma.

METHODS

We used a retrospective cohort study design with 12-month baseline and follow-up periods. We identified a cohort of older adults (age ≥ 66 years) diagnosed with breast, prostate, colorectal cancers, or Non-Hodgkin's lymphoma between January 2014 and December 2014. We assessed low-value care and patient out-of-pocket expenditure in the follow-up period. We identified relevant low-value services using ICD9/ICD10 and CPT/HCPCS codes from the linked health claims and patient out-of-pocket expenditure from Medicare claim files and expressed expenditure in 2016 USD.

RESULTS

About 29 % of older adults received at least one low-value care procedure during the follow-up period. Low-value care differed by gender, and rates were higher in women with colorectal cancer (32.7 %) vs. (28.8 %) and NHL (40 %) vs. (39 %) compared to men. Individuals who received one or more low-value care procedures had significantly higher mean out-of-pocket expenditure ($8,726 ± $7,214) vs. ($6,802 ± $6,102). XGBOOST, a machine learning algorithm revealed that low-value care was among the five leading predictors of OOP expenditure.

CONCLUSION

One in four older adults with incident cancer received low-value care in 12-months after a cancer diagnosis. Across all cancer populations, individuals who received low-value care had significantly higher out-of-pocket expenditure. Excess out-of-pocket expenditure was driven by low-value care, fragmentation of care, and an increasing number of pre-existing chronic conditions.

POLICY STATEMENT

This study focuses on health policy issues, specifically value-based care and its findings have important clinical and policy implications for Centers for Medicare and Medicaid Services (CMS) which has issued a roadmap for states to accelerate the adoption of value-based care, with the Department of Health and Human Services (HHS) setting a goal of converting 50 % of traditional Medicare payment systems to alternative payment models tied to value-based care by 2022.