Our algorithms rely exclusively on relevant clinical parameters. We don't employ big data techniques to draw loose inferences from codes or claims or require structured data. Our machine learning tools are embedded in our data extraction platform rather than in our algorithms. They only operate unsupervised at thresholds that yield 100% accuracy. We manage uncertainty with technology-enabled abstraction by trained healthcare professionals.
This allows us to build tools that are unrivalled in their precision. It's why our ATTR-CM algorithm is driven by reported echo parameters like IVSD, E/E' and GLS, why our ALS algorithm can incorporate the presence of sharp waves on EMG in patients with clinical fasciculations and why our IL-1 mediated autoinflammatory diseases algorithm can account for fever periodicity in its calculations.
Our reports clearly and transparently explain why patients may be suitable for follow-up investigations or referrals to specialty centres. The clinical parameters that drove our calculations are always reported alongside pertinent positives, pertinent negatives and other signs or symptoms necessary to complete the clinical picture.
Our mission is to empower physicians with better tools. Our reports reflect our best efforts to realize it.
We deploy our digital diagnostics exclusively at our clinical data labs because:
It allows us to control for variability in source data formats;
We can achieve greater reliability and accuracy than if we were to deploy our solutions locally; and
It makes life easier for our customers.
We go to great lengths to keep health information safe and secure including through the adoption of novel asymmetric de-identification and encryption techniques.