Dynamic multi-outcome prediction after injury: Applying adaptive machine learning for precision medicine in trauma.

Dynamic multi-outcome prediction after injury: Applying adaptive machine learning for precision medicine in trauma.

Publication date: Jul 09, 2018

Machine learning techniques have demonstrated superior discrimination compared to conventional statistical approaches in predicting trauma death. The objective of this study is to evaluate whether machine learning algorithms can be used to assess risk and dynamically identify patient-specific modifiable factors critical to patient trajectory for multiple key outcomes after severe injury.

SuperLearner, an ensemble machine-learning algorithm, was applied to prospective observational cohort data from 1494 critically-injured patients. Over 1000 agnostic predictors were used to generate prediction models from multiple candidate learners for outcomes of interest at serial time points post-injury. Model accuracy was estimated using cross-validation and area under the curve was compared to select among predictors. Clinical variables responsible for driving outcomes were estimated at each time point.

SuperLearner fits demonstrated excellent cross-validated prediction of death (overall AUC 0.94-0.97), multi-organ failure (overall AUC 0.84-0.90), and transfusion (overall AUC 0.87-0.9) across multiple post-injury time points, and good prediction of Acute Respiratory Distress Syndrome (overall AUC 0.84-0.89) and venous thromboembolism (overall AUC 0.73-0.83). Outcomes with inferior data quality included coagulopathic trajectory (AUC 0.48-0.88). Key clinical predictors evolved over the post-injury timecourse and included both anticipated and unexpected variables. Non-random missingness of data was identified as a predictor of multiple outcomes over time.

Machine learning algorithms can be used to generate dynamic prediction after injury while avoiding the risk of over- and under-fitting inherent in ad hoc statistical approaches. SuperLearner prediction after injury demonstrates promise as an adaptable means of helping clinicians integrate voluminous, evolving data on severely-injured patients into real-time, dynamic decision-making support.

Open Access PDF

Christie, S.A., Conroy, A.S., Callcut, R.A., Hubbard, A.E., and Cohen, M.J. Dynamic multi-outcome prediction after injury: Applying adaptive machine learning for precision medicine in trauma. 04398. 2018 PLoS One (14):4.

Concepts Keywords
Agnostic Machine learning
AUC Branches of biology
Cohort Medicine
Cross Sports injury
Cross Validation Major trauma
Discrimination Prediction
Machine Learning Machine learning
Machine Learning Algorithm Learning
Organ Cybernetics
Transfusion Medical specialties
Trauma Post injury
Venous Thromboembolism Venous thromboembolism
Dynamic injury
Injury

Semantics

Type Source Name
gene UNIPROT ASRGL1
gene UNIPROT ABCB6
drug DRUGBANK Abacavir
drug DRUGBANK Tropicamide
drug DRUGBANK (S)-Des-Me-Ampa
gene UNIPROT GNPTAB
gene UNIPROT SLC26A5
gene UNIPROT MARK1
gene UNIPROT F11R
gene UNIPROT SHPK
disease MESH Shock
disease MESH ventilator associated pneumonia
gene UNIPROT KCNJ11
gene UNIPROT INTU
gene UNIPROT DUOXA1
gene UNIPROT APPL1
gene UNIPROT SOAT1
gene UNIPROT TNIP1
drug DRUGBANK MCC
gene UNIPROT SKAP2
drug DRUGBANK Heparin
disease DOID CCM
gene UNIPROT COL9A3
gene UNIPROT COMP
gene UNIPROT COL9A1
gene UNIPROT COL9A2
gene UNIPROT SCN8A
gene UNIPROT ELL
gene UNIPROT ELAVL2
drug DRUGBANK Huperzine B
gene UNIPROT RELA
gene UNIPROT LAT2
drug DRUGBANK Etoperidone
drug DRUGBANK Ethanol
drug DRUGBANK Fenamole
disease DOID ers
gene UNIPROT CCHCR1
gene UNIPROT SHBG
gene UNIPROT PPT1
gene UNIPROT PPP5C
drug DRUGBANK Podofilox
gene UNIPROT DYRK3
gene UNIPROT IK
gene UNIPROT EGR3
gene UNIPROT THOP1
disease MESH Embolism
gene UNIPROT FBN1
gene UNIPROT KAT8
gene UNIPROT AMH
gene UNIPROT NPL
drug DRUGBANK Zolmitriptan
gene UNIPROT ADI1
drug DRUGBANK Saquinavir
drug DRUGBANK Prothrombin
gene UNIPROT SET
gene UNIPROT ATM
disease DOID ARDS
disease MESH nosocomial infection
disease MESH Multiple Organ Failure
disease MESH Coma
gene UNIPROT EHD1
disease MESH emergency
disease MESH Inflammation
gene UNIPROT LARGE1
gene UNIPROT SPINK5
gene UNIPROT LITAF
gene UNIPROT DEPP1
gene UNIPROT GOPC
gene UNIPROT SLC14A1
gene UNIPROT ERAL1
gene UNIPROT ESR1
disease MESH critically ill
gene UNIPROT ALPK3
gene UNIPROT MAK
gene UNIPROT CEP55
gene UNIPROT TNFSF13
gene UNIPROT ANP32B
disease DOID Syndrome
disease MESH Syndrome
gene UNIPROT NAA50
disease MESH venous thromboembolism
disease MESH Acute Respiratory Distress Syndrome
gene UNIPROT SLC35G1
disease MESH multiple
disease MESH death
disease MESH multi

Original Article

Leave a Comment

Your email address will not be published. Required fields are marked *