Model-based and Model-free Machine Learning Techniques for Diagnostic Prediction and Classification of Clinical Outcomes in Parkinson’s Disease

Abstract

In this study, we apply a multidisciplinary approach to investigate falls in PD patients using clinical, demographic and neuroimaging data from two independent initiatives (University of Michigan and Tel Aviv Sourasky Medical Center). Using machine learning techniques, we construct predictive models to discriminate fallers and non-fallers. Through controlled feature selection, we identified the most salient predictors of patient falls including gait speed, Hoehn and Yahr stage, postural instability and gait difficulty-related measurements. The model-based and model-free analytical methods we employed included logistic regression, random forests, support vector machines, and XGboost. The reliability of the forecasts was assessed by internal statistical (5-fold) cross validation as well as by external out-of-bag validation. Four specific challenges were addressed in the study: Challenge 1, develop a protocol for harmonizing and aggregating complex, multisource, and multi-site Parkinson’s disease data; Challenge 2, identify salient predictive features associated with specific clinical traits, e.g., patient falls; Challenge 3, forecast patient falls and evaluate the classification performance; and Challenge 4, predict tremor dominance (TD) vs. posture instability and gait difficulty (PIGD). Our findings suggest that, compared to other approaches, model-free machine learning based techniques provide a more reliable clinical outcome forecasting of falls in Parkinson’s patients, for example, with a classification accuracy of about 70–80%.

Machine learning methods for optimal prediction of motor outcome in Parkinson’s disease

“…Novel biomarkers may enable more accurate stratification of PD based on the expected prognosis. There is also a significant interest in using the biomarkers for prediction of disease outcome [20][21][22][23][24][25]28], to properly adapt clinical trial studies as applied to appropriate patients.…”

A comparative study of machine learning algorithms for predicting acute kidney injury after liver cancer resection

“…This can induce hepatic ischemiareperfusion injury. It can cause not only liver dysfunction, but also kidney injury (Sheridan et al, 2016;Gao et al, 2018). At the same time, due to surgical trauma, decreased blood flow in the liver and decreased kidney circulation, granulocyte elastase release and other factors, postoperative renal damage can also occur (Fonseca-NetoI et al, 2012).…”

Deep Knockoffs

“…The performance of the deep knockoff machine is typically not very sensitive to this choice, although we will discuss how different ratios work better with certain distributions. Upon completion of training, the goodness-of-fit of the machines is quantified in terms of the metrics defined in Section 5.1, namely the matching of second moments (14), the maximum mean discrepancy score (15), the k-nearest neighbors test (16) with k = 1 and the energy test (18). These measures are evaluated on knockoff copies generated for 1000 previously unseen independent samples drawn from the same distribution P X .…”

“…For this purpose, almost any available method from statistics and machine learning can be applied to the vector of labels Y and the augmented data matrix [X,X] ∈ R n×2p , with the only fundamental rule that the original variables and the knockoffs should be treated equally; this is saying that the method should not use any information revealing which variable is a knockoff and which is not. Examples include sparse generalized linear models [1,3], random forests [15], support vector machines and deep neural networks [9,10]. Each pair of Z j andZ j is then combined through an antisymmetric function into the statistics W j , e.g.…”