Statistics versus machine learning

Bzdok, Danilo; Altman, Naomi; Krzywinski, Martin

doi:10.1038/nmeth.4642

Cited by 991 publications

(776 citation statements)

References 8 publications

Supporting

Mentioning

688

Contrasting

Unclassified

Order By: Relevance

“…Inference and prediction are two sides of a coin when inquiring human health and disease (1)(2)(3). Let's take diabetes mellitus as a motivating example.…”

Section: Introductionmentioning

confidence: 99%

Prediction and inference diverge in biomedicine: Simulations and real-world data

Bzdok¹,

Engemann²,

Grisel³

et al. 2018

Preprint

Self Cite

View full text Add to dashboard Cite

In the 20 th century many advances in biological knowledge and evidence-based medicine were supported by p-values and accompanying methods. In the beginning 21 st century, ambitions towards precision medicine put a premium on detailed predictions for single individuals. The shift causes tension between traditional methods used to infer statistically significant group differences and burgeoning machine-learning tools suited to forecast an individual's future. This comparison applies the linear model for identifying significant contributing variables and for finding the most predictive variable sets. In systematic data simulations and common medical datasets, we explored how statistical inference and pattern recognition can agree and diverge. Across analysis scenarios, even small predictive performances typically coincided with finding underlying significant statistical relationships. However, even statistically strong findings with very low p-values shed little light on their value for achieving accurate prediction in the same dataset. More complete understanding of different ways to define 'important' associations is a prerequisite for reproducible research findings that can serve to personalize clinical care. Keywords: scientific discovery | data science | variable importance | intelligent algorithms | reproducibility'Change your statistical philosophy and all of a sudden different things become important'

show abstract

“…Inference and prediction are two sides of a coin when inquiring human health and disease (1)(2)(3). Let's take diabetes mellitus as a motivating example.…”

Section: Introductionmentioning

confidence: 99%

Prediction and inference diverge in biomedicine: Simulations and real-world data

Bzdok¹,

Engemann²,

Grisel³

et al. 2018

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…10 The feasibility of ML has been demonstrated previously in the CAD risk stratification field. 7 Moreover, these alterations were associated with reductions in occurrence of non-fatal MI's.…”

Section: Machine Learning To Improve Integration Of Coronary Plaque Amentioning

confidence: 89%

“…3,8,9 Plaque information derived during CCTA acquisition and subsequently classified according to the 16-segment coronary tree model is typically integrated into a single score, assuming linear relationships between CAD extent and risk. 10 Machine learning (ML) is a field in computer science that uses algorithms to combine a big data in order to optimize prediction. Previous studies have demonstrated that ML can increase predictive value for death and myocardial ischemia compared to conventional scores.…”

Section: Introductionmentioning

confidence: 99%

Maximization of the usage of coronary CTA derived plaque information using a machine learning based algorithm to improve risk stratification; insights from the CONFIRM registry

Rosendael

Maliakal

Kolli

et al. 2018

Journal of Cardiovascular Computed Tomography

147

View full text Add to dashboard Cite

“…Many statistical methods are limited in their ability to sort through large numbers of predictors (14). Data mining using machine learning, which is particularly well suited for identifying predictive factors among thousands of variables (15,16), has successfully identified predictor variables for a diverse set of outcomes (17)(18)(19)(20)(21).…”

Section: Introductionmentioning

confidence: 99%

Factors associated with opioid cessation: a machine learning approach

Cox

Sherva

Lunetta

et al. 2019

Preprint

View full text Add to dashboard Cite

Background and Aims People with opioid use disorder (OUD) can stop using opioids on their own, with help from groups and with treatment, but there is limited research on the factors that influence opioid cessation. Methods We employed multiple machine learning prediction algorithms (LASSO, random forest, deep neural network, and support vector machine) to assess factors associated with ceasing opioid use in a sample comprised of African Americans (AAs) and European Americans (EAs) who met DSM-5 criteria for mild to severe OUD. Values for several thousand demographic, alcohol and other drug use, general health, and behavioral variables, as well as diagnoses for other psychiatric disorders, were obtained for each participant from a detailed semi-structured interview. Results Support vector machine models performed marginally better on average than those derived using other machine learning methods with maximum prediction accuracies of 75.4% in AAs and 79.4% in EAs. Subsequent stepwise regression analyses that considered the 83 most highly ranked variables across all methods and models identified less recent cocaine use (p<5×10 -8 ), a shorter duration of opioid use (p<5×10 -6 ), and older age (p<5×10 -9 ) as the strongest independent predictors of opioid cessation. Factors related to drug use comprised about half of the significant independent predictors, with other predictors related to non-drug use behaviors, psychiatric disorders, overall health, and demographics. Conclusions These proof-ofconcept findings provide information that can help develop strategies for improving OUD management and the methods we applied provide a framework for personalizing OUD treatment.

show abstract

Statistics versus machine learning

Cited by 991 publications

References 8 publications

Prediction and inference diverge in biomedicine: Simulations and real-world data

Prediction and inference diverge in biomedicine: Simulations and real-world data

Maximization of the usage of coronary CTA derived plaque information using a machine learning based algorithm to improve risk stratification; insights from the CONFIRM registry

Factors associated with opioid cessation: a machine learning approach

Contact Info

Product

Resources

About