Algorithms in Medicine
Algorithms have a growing place in medicine. There are at least two kinds of algorithms – practice guidelines and predictive.
Algorithms have a growing place in medicine. There are at least two kinds of algorithms – practice guidelines and predictive.
Practice guideline or best practice algorithms may emerge out of practice-based research or consensus panels which recommend specific practices to improve care. One example relates to algorithms in intensive care. Cleveland Clinic has excellent examples on their medical education website which include everything from weaning to acid-based disorders. Some may criticize this as “cookbook medicine.” But with the complexity of care, especially intensive care, these kind of algorithms should be welcome. How these can be integrated into clinical decision support in electronic medical records is yet to be fully realized.
Predictive models, on the other hand, use patient-specific data to determine future risk of disease or risk. Predictive models are usually based upon large data sets and in recent years that means EMR data. A couple of examples:
- Predicting 6-Year Mortality Risk in Patients With Type 2 Diabetesby one of my colleagues
- Development and Validation of a Predictive Algorithm to Identify Adult Asthmatics from Medical Services and Pharmacy Claims Databases
One outcome of predictive models is risk calculators or nomograms. Several examples can be found here:
- http://prostatecancerinfolink.net/tips-tools/kattan-nomograms/ by another colleague
- nomograms.org
Algorithms can potentially provide real value in medical practice and to patient decision making as well. They are part of a movement toward artificial intelligence in health care.
