Erroneous data and drug industry bias can impair machine learning algorithmsBMJ 2019; 367 doi: https://doi.org/10.1136/bmj.l6042 (Published 18 October 2019) Cite this as: BMJ 2019;367:l6042
- Thomas Birk Kristiansen, medical doctor
- Ishøjcentrets Læger, Ishøj Store Torv 22, 2635 Ishøj, Denmark
This important analysis focuses on the major obstacles that preventive precision medicine must overcome to fulfil its potential.1 Two further obstacles should be brought to attention.
First, erroneous data have great implications for machine learning algorithms used in predictive precision medicine. This holds true for historical data used to train algorithms and for new data used to make …