Professor Gill Bejerano of the University of Stanford has developed an algorithm to facilitate the diagnosis of rare diseases. The algorithm compares patient symptoms and genetic data to a database. In this way it accelerates the diagnosis and allows to identify even rare and little known diseases.
Professor Bejerano and colleagues describe the algorithm in an article published in Genetics in Medicine. The code automates the most mechanical part of the diagnostic work, that is the comparison between the patient's genetic sequences and the scientific literature. Without the help of the computer, the process takes about 20-40 hours per patient. The algorithm - Phrank - cuts 90% of the time needed.
Phrank compares the patient's symptoms and genes with data from the medical literature. Unlike other algorithms of this kind, Phrank is not bound to a specific database. After the comparison, the algorithm generates a list of possible genetic diseases, assigning each a probability score. In this way the doctor has a starting point to make a diagnosis, all in less than an hour.
The team of researchers validated Phrank on the genetic and medical data of 169 patients. The algorithm proved to be much more effective than all those elaborated up to now. However, it has been tested on artificial patients: for such studies, it is difficult to find a sufficient number of real patients, at least in the early stages.
To confirm the effectiveness of Phrank, real patient data will be required. With all its limitations, however, it is a good starting point to facilitate the work of doctors.