Area 1: Add COVID-19 to the differential diagnoses
Currently, the CA does not include the differential diagnosis with COVID-19, as all the patients presenting with suggestive symptoms undergo a SARS-CoV-2 testing before further diagnostic pathway. As soon as the epidemiological scenario would allow, the SARS-CoV-2 will be added to the CA.Area 2: Collapse signs and symptoms
The signs and symptoms of lower respiratory tract infection have little predictive value. Current version of the heuristic algorithm treats all point of care tests (POCTs) and their outcomes as independent. The signs and symptoms should be collapsed into a general predictor that would compute a probability distribution for a presence/absence vector across a comprehensive list of symptoms .Area 3: Precise biomarker values
The current version of the algorithm includes biomarker-based tests in binary forms using the different cutoffs found in the meta-analysis. However, provided enough patient-level data, the regressor models could be trained to replace the binary tests, and provide a more accurate estimates.Area 4: Tailoring by the local epidemiological conditions
The formulation of the CA enables utilization of the precise prevalence distribution based on global surveillance data. The CA could be recomputed automatically at local, national level, updating with the current prevalence data. Ideally, it could be directly connected to regional surveillance statistics.Area 5: Tailoring to the POCTs availability and cost
The formulation of the CA enables precise per-patient computation of the cost of entire diagnostic pathway, providing the real costs of the available POCTs. The next update of the CAs could include option to select a cheaper algorithm among the equally-well performing algorithms.Area 6: Bringing to live underlying meta-analysis
The underlying meta-analysis did not enable us to subgroup the patients by the population characteristics: demographics, comorbidities or vaccination status. We plan to automatize the screening step of the meta-analysis. A server will automatically assess the new evidence and alert the reviewer in selected time intervals about the articles matching the criteria. Applying this methodology, the meta-analysis could be easily updated, and with enough subgroup representation, we would be able to develop population-specific CAs.Area 7: Validation
So far, the CA were validated with the patient-level data from different studies. External validation (and translation to other languages) is currently under evaluation.The CA will be updated with the results of the Value-Dx clinical trials as soon as completed. Next literature update is under discussion.