Supervised learning applied to classifying fallers versus non-fallers among older adults with cancer.

INTRODUCTION: Supervised machine learning approaches are increasingly used to analyze clinical data, including in geriatric oncology. This study presents a machine learning approach to understand falls in a cohort of older adults with advanced cancer starting chemotherapy, including fall prediction and identification of contributing factors. MATERIALS AND METHODS: This secondary analysis of prospectively collected data from the GAP 70+ Trial (NCT02054741; PI: Mohile) enrolled patients aged ≥70 with advanced cancer and ≥ 1 geriatric assessment domain impairment who planned to start a new cancer treatment regimen. Of ≥2000 baseline variables ("features") collected, 73 were selected based on clinical judgment. Machine learning models to predict falls at three months were developed, optimized, and tested using data from 522 patients. A custom data preprocessing pipeline was implemented to prepare data for analysis. Both undersampling and oversampling techniques were applied to balance the outcome measure. Ensemble feature selection was applied to identify and select the most relevant features. Four models (logistic regression [LR], k-nearest neighbor [kNN], random forest [RF], and MultiLayer Perceptron [MLP]) were trained and subsequently tested on a holdout set. Receiver operating characteristic (ROC) curves were generated and area under the curve (AUC) was calculated for each model. SHapley Additive exPlanations (SHAP) values were utilized to further understand individual feature contributions to observed predictions. RESULTS: Based on the ensemble feature selection algorithm, the top eight features were selected for inclusion in the final models. Selected features aligned with clinical intuition and prior literature. The LR, kNN, and RF models performed equivalently well in predicting falls in the test set, with AUC values 0.66-0.67, and the MLP model showed AUC 0.75. Ensemble feature selection resulted in improved AUC values compared to using LASSO alone. SHAP values, a model-agnostic technique, revealed logical associations between selected features and model predictions. DISCUSSION: Machine learning techniques can augment hypothesis-driven research, including in older adults for whom randomized trial data are limited. Interpretable machine learning is particularly important, as understanding which features impact predictions is a critical aspect of decision-making and intervention. Clinicians should understand the philosophy, strengths, and limitations of a machine learning approach applied to patient data.

Plasma MIA, CRP, and Albumin Predict Cognitive Decline in Parkinson's Disease.

OBJECTIVE: Using a multi-cohort, discovery-replication-validation design, we sought new plasma biomarkers that predict which individuals with Parkinson's disease (PD) will experience cognitive decline. METHODS: In 108 discovery cohort PD individuals and 83 replication cohort PD individuals, we measured 940 plasma proteins on an aptamer-based platform. Using proteins associated with subsequent cognitive decline in both cohorts, we trained a logistic regression model to predict which patients with PD showed fast (> = 1 point drop/year on Montreal Cognitive Assessment [MoCA]) versus slow (< 1 point drop/year on MoCA) cognitive decline in the discovery cohort, testing it in the replication cohort. We developed alternate assays for the top 3 proteins and confirmed their ability to predict cognitive decline - defined by change in MoCA or development of incident mild cognitive impairment (MCI) or dementia - in a validation cohort of 118 individuals with PD. We investigated the top plasma biomarker for causal influence by Mendelian randomization (MR). RESULTS: A model with only 3 proteins (melanoma inhibitory activity protein [MIA], C-reactive protein [CRP], and albumin) separated fast versus slow cognitive decline subgroups with an area under the curve (AUC) of 0.80 in the validation cohort. The individuals with PD in the validation cohort in the top quartile of risk for cognitive decline based on this model were 4.4 times more likely to develop incident MCI or dementia than those in the lowest quartile. Genotypes at MIA single nucleotide polymorphism (SNP) rs2233154 associated with MIA levels and cognitive decline, providing evidence for MIA's causal influence. CONCLUSIONS: An easily obtained plasma-based predictor identifies individuals with PD at risk for cognitive decline. MIA may participate causally in development of cognitive decline. ANN NEUROL 2022;92:255-269.

Neurofilament Light Chain as a Biomarker for Cognitive Decline in Parkinson Disease.

BACKGROUND: Neurofilament light chain protein (NfL) is a promising biomarker of neurodegeneration. OBJECTIVES: To determine whether plasma and CSF NfL (1) associate with motor or cognitive status in Parkinson's disease (PD) and (2) predict future motor or cognitive decline in PD. METHODS: Six hundred and fifteen participants with neurodegenerative diseases, including 152 PD and 200 healthy control participants, provided a plasma and/or cerebrospinal fluid (CSF) NfL sample. Diagnostic groups were compared using the Kruskal-Wallis rank test. Within PD, cross-sectional associations between NfL and Unified Parkinson's Disease Rating Scale Part III (UPDRS-III) and Mattis Dementia Rating Scale (DRS-2) scores were assessed by linear regression; longitudinal analyses were performed using linear mixed-effects models and Cox regression. RESULTS: Plasma and CSF NfL levels correlated substantially (Spearman r = 0.64, P < 0.001); NfL was highest in neurocognitive disorders. PD participants with high plasma NfL were more likely to develop incident cognitive impairment (HR 5.34, P = 0.005). CONCLUSIONS: Plasma NfL is a useful prognostic biomarker for PD, predicting clinical conversion to mild cognitive impairment or dementia. © 2021 International Parkinson and Movement Disorder Society.

Resident physician exposure to novel coronavirus (2019-nCoV, SARS-CoV-2) within New York City during exponential phase of COVID-19 pandemic: Report of the New York City Residency Program Directors COVID-19 Research Group.

Background From March 2-April 12, 2020, New York City (NYC) experienced exponential growth of the COVID-19 pandemic due to novel coronavirus (SARS-CoV-2). Little is known regarding how physicians have been affected. We aimed to characterize COVID-19 impact on NYC resident physicians. Methods IRB-exempt and expedited cross-sectional analysis through survey to NYC residency program directors (PDs) April 3-12, 2020, encompassing events from March 2-April 12, 2020. Findings From an estimated 340 residency programs around NYC, recruitment yielded 91 responses, representing 24 specialties and 2,306 residents. 45.1% of programs reported at least one resident with confirmed COVID-19: 101 resident physicians were confirmed COVID-19-positive, with additional 163 residents presumed positive for COVID-19 based on symptoms but awaiting or unable to obtain testing. 56.5% of programs had a resident waiting for, or unable to obtain, COVID-19 testing. Two COVID-19-positive residents were hospitalized, with one in intensive care. Among specialties with >100 residents represented, negative binomial regression indicated that infection risk differed by specialty (p=0.039). Although most programs (80%) reported quarantining a resident, with 16.8% of residents experiencing quarantine, 14.9% of COVID-19-positive residents were not quarantined. 90 programs, encompassing 99.2% of the resident physicians, reported reuse or extended mask use, and 43 programs, encompassing 60.4% of residents, felt that personal protective equipment (PPE) was suboptimal. 65 programs (74.7%) have redeployed residents elsewhere to support COVID-19 efforts. Interpretation Many resident physicians around NYC have been affected by COVID-19 through direct infection, quarantine, or redeployment. Lack of access to testing and concern regarding suboptimal PPE are common among residency programs. Infection risk may differ by specialty. Funding AHA, MPB, RWSC, CGM, LRDG, and JDH are supported by NEI Core Grant P30EY019007, and unrestricted grant from RPB. ACP and JS are supported by Parker Family Chair. SXX is supported by University of Pennsylvania.

New York City COVID-19 resident physician exposure during exponential phase of pandemic.

BACKGROUNDFrom March 2, 2020, to April 12, 2020, New York City (NYC) experienced exponential growth of the COVID-19 pandemic due to novel coronavirus (SARS-CoV-2). Little is known regarding how physicians have been affected. We aimed to characterize the COVID-19 impact on NYC resident physicians.METHODSIRB-exempt and expedited cross-sectional analysis through survey to NYC residency program directors April 3-12, 2020, encompassing events from March 2, 2020, to April 12, 2020.RESULTSFrom an estimated 340 residency programs around NYC, recruitment yielded 91 responses, representing 24 specialties and 2306 residents. In 45.1% of programs, at least 1 resident with confirmed COVID-19 was reported. One hundred one resident physicians were confirmed COVID-19-positive, with an additional 163 residents presumed positive for COVID-19 based on symptoms but awaiting or unable to obtain testing. Two COVID-19-positive residents were hospitalized, with 1 in intensive care. Among specialties with more than 100 residents represented, negative binomial regression indicated that infection risk differed by specialty (P = 0.039). In 80% of programs, quarantining a resident was reported. Ninety of 91 programs reported reuse or extended mask use, and 43 programs reported that personal protective equipment (PPE) was suboptimal. Sixty-five programs (74.7%) redeployed residents elsewhere to support COVID-19 efforts.CONCLUSIONMany resident physicians around NYC have been affected by COVID-19 through direct infection, quarantine, or redeployment. Lack of access to testing and concern regarding suboptimal PPE are common among residency programs. Infection risk may differ by specialty.FUNDINGNational Eye Institute Core Grant P30EY019007; Research to Prevent Blindness Unrestricted Grant; Parker Family Chair; University of Pennsylvania.