Development and Validation of a Non-Invasive, Chairside Oral Cavity Cancer Risk Assessment Prototype Using Machine Learning Approach

Oral cavity cancer (OCC) is associated with high morbidity and mortality rates when diagnosed at late stages. Early detection of increased risk provides an opportunity for implementing prevention strategies surrounding modifiable risk factors and screening to promote early detection and intervention...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Journal of personalized medicine Ročník 12; číslo 4; s. 614
Hlavní autori: Shimpi, Neel, Glurich, Ingrid, Rostami, Reihaneh, Hegde, Harshad, Olson, Brent, Acharya, Amit
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Switzerland MDPI AG 11.04.2022
MDPI
Predmet:
60 APPLIED LIFE SCIENCES
Alcohol
Algorithms
Cancer therapies
Clinics
Data mining
Data models
Data warehouses
Datasets
Electronic medical records
Head & neck cancer
Hispanic Americans
Informatics
Learning algorithms
Machine learning
Morbidity
Mortality
Oral cancer
Oral carcinoma
Oral cavity
patient care management
Patients
Phenotypes
Precision medicine
Primary care
Risk assessment
Risk factors
Tobacco
Tumors
risk assessment
precision medicine
machine learning
patient care management
oral cancer
ISSN:2075-4426, 2075-4426
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:Oral cavity cancer (OCC) is associated with high morbidity and mortality rates when diagnosed at late stages. Early detection of increased risk provides an opportunity for implementing prevention strategies surrounding modifiable risk factors and screening to promote early detection and intervention. Historical evidence identified a gap in the training of primary care providers (PCPs) surrounding the examination of the oral cavity. The absence of clinically applicable analytical tools to identify patients with high-risk OCC phenotypes at point-of-care (POC) causes missed opportunities for implementing patient-specific interventional strategies. This study developed an OCC risk assessment tool prototype by applying machine learning (ML) approaches to a rich retrospectively collected data set abstracted from a clinical enterprise data warehouse. We compared the performance of six ML classifiers by applying the 10-fold cross-validation approach. Accuracy, recall, precision, specificity, area under the receiver operating characteristic curve, and recall–precision curves for the derived voting algorithm were: 78%, 64%, 88%, 92%, 0.83, and 0.81, respectively. The performance of two classifiers, multilayer perceptron and AdaBoost, closely mirrored the voting algorithm. Integration of the OCC risk assessment tool developed by clinical informatics application into an electronic health record as a clinical decision support tool can assist PCPs in targeting at-risk patients for personalized interventional care.
Bibliografia:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
USDOE
National Institutes of Health (NIH)
AC02-05CH11231; UL1TR000427
ISSN:2075-4426
2075-4426
DOI:10.3390/jpm12040614