This is from a CDx Life email
Data Analysis and Sensor Algorithm Development
Our current strain match software utilizes a proprietary algorithm that leverages Support Vector Machine (SVM) methodology. These are supervised learning models with associated learning algorithms that analyze data and recognize patterns, used for classification and regression analysis. With the Data that AZ Med (our partner testing lab) populates, the text documents are taken and input into a directory. The software processes the data in the text documents and passes it through the SVM algorithm. The algorithm predicts the strain based upon learned training sets and outputs the processed data identifying it's strain name. If no strain is found, the software will indicate an unidentified match.
Based upon the current database (over 250 samples tested), and the current algorithm developed, we can identify and match strains with an accuracy of 90%. Meaning out of 100 samples that will be analyzed via a MyDx unit, 90 are accurately identified. As we continue to populate the database, we can continue conditioning and monitoring this accuracy. The many more samples we plan on testing before December will drive this number closer to 100% accuracy for strain match. We will reach this goal by by increasing the sample size and number of scans per sample.
Here's an interview with the CEO