What are your primary sources of data?
STAT Trials Pulse leverages data from ClinicalTrials.gov which is processed and updated to ensure that it is always current. We highlight when the data content is modified, and also provide a complete description of those modifications. The clinical trials data currently used in STAT Trials Pulse is exclusively from interventional studies.
This platform also uses publicly available data courtesy of the U.S. National Library of Medicine (NLM), the National Institutes of Health, and the Department of Health and Human Services. NLM is not responsible for the product and does not endorse or recommend this or any other product.
We also monitor and incorporate data from press releases from thousands of companies made available to use via the Associated Press.
Disclaimer: Applied XL does not warrant or assume any legal liability or responsibility for the accuracy, completeness, or usefulness of any information or data disclosed.
How do you guarantee the quality of your algorithms?
All of our algorithms are vetted and validated by humans who are experts in their fields. Exploration of data is performed by using computational journalism methods to help us identify any eventual errors, biases, and areas for further refinement.
During model training, we ensure that enough data has been provided to the algorithm to account for unusual or edge cases. A network of curated experts, including STAT journalists and researchers, help us evaluate outliers that are flagged by our system. This feedback is used to recalibrate models, or to verify the accuracy of important events. The human-in-the-loop technology integrated in our platform allows for continuous feedback that improves our algorithms on an ongoing basis.
Classifications that are algorithmically generated always include a disclosure to provide transparency and visibility into the process. These disclosures also offer the opportunity for the user to provide feedback so we can reassess our current models. Experts and users can contribute with this discrete data feedback, which is used to retain our machine learning algorithms to achieve increased levels of precision and to help control data drift. We leverage strategic human input to surface new labels that did not previously exist, ensuring a dynamic and editorially sound classification model.
If you'd like a guided demo of the platform with our team, you can pick a slot on our calendar.
If you need support or would like to provide feedback, reach out to us at support@appliedxl.com.