Rxivist logo

A comparative study of machine learning algorithms in predicting severe complication after bariatric surgery

By Yang Cao, Xin Fang, Johan Ottosson, Erik Näslund, Erik Stenberg

Posted 27 Jul 2018
bioRxiv DOI: 10.1101/376038 (published DOI: 10.3390/jcm8050668)

Accurate models to predict severe postoperative complications could be of value in the preoperative assessment of potential candidates for bariatric surgery. Traditional statistical methods have so far failed to produce high accuracy. To find a useful algorithm to predict the risk for severe complication after bariatric surgery, we trained and compared 29 supervised machine learning (ML) algorithms using information from 37,811 patients operated with a bariatric surgical procedure between 2010 and 2014 in Sweden. The algorithms were then tested on 6,250 patients operated in 2015. Most ML algorithms showed high accuracy (>90%) and specificity (>0.9) in both the training and test data. However, none achieved an acceptable sensitivity in the test data. ML methods may improve accuracy of prediction but we did not yet identify one with a high enough sensitivity that can be used in clinical praxis in bariatric surgery. Further investigation on deeper neural network algorithms is needed.

Download data

  • Downloaded 765 times
  • Download rankings, all-time:
    • Site-wide: 35,675
    • In epidemiology: 1,812
  • Year to date:
    • Site-wide: 105,724
  • Since beginning of last month:
    • Site-wide: 49,618

Altmetric data

Downloads over time

Distribution of downloads per paper, site-wide