#USA #Cambridge #AI Technology #TriNetX #Clinical Notes

TriNetX Launches Advanced AI-Powered Fact Extraction

As artificial intelligence (AI) continues to advance in healthcare, the ability to extract meaningful insights from clinical notes has reached new heights. TriNetX®, a global leader in federated health data, recently introduced a revolutionary capability called NLP-extracted Facts from Clinical Notes, available through its TriNetX LIVE™ platform and downloadable datasets. This innovative approach provides researchers with analysis-ready data points extracted from over 800 million clinical notes, a figure expected to soar to around 1 billion by the end of 2026.

Clinical notes are troves of valuable information, including critical elements such as disease stage, performance status, and biomarker results. However, extracting this data in a reliable and meaningful manner has been a significant challenge. TriNetX marries AI with extensive expertise in clinical informatics to create a robust pipeline that ensures high-quality data extraction. The company employs a human-in-the-loop approach, training AI models on a extensive set of annotated gold standard notes. Unlike methods that solely focus on data extraction, TriNetX has implemented multiple safeguards for every fact it reports:

1. Human Oversight: Clinical experts thoroughly review AI-generated tags, and those corrections are used to retrain models. Each module is required to achieve a precision score of 0.90 or higher when compared to human annotations.
2. Clinical Corroboration: Extracted data is validated against expected clinical patterns, enhancing trustworthiness. For instance, survival analysis based on Eastern Cooperative Oncology Group (ECOG) performance status confirmed poor survival rates for patients with scores of 2 or higher.
3. Continuous Monitoring: TriNetX continuously tracks the performance of models as they adapt to new healthcare organizations and note types, fine-tuning them to reflect any variations.

Early results have demonstrated the efficacy of these safeguards. Notably, the inclusion of notes-derived facts improved the coverage of smoking status from 22% to 67%. Furthermore, the approach yielded six times more patients than traditional structured data in a lung cancer cohort study, revealing ECOG and Karnofsky performance statuses for 292,000 oncology patients without that information being present in structured fields.

According to TriNetX CEO August Calhoun, the company's meticulous methodology is what distinguishes them from others in the industry. “Pointing AI at a chart is simple, but TriNetX sets itself apart by applying scientific rigor and human expertise in a comprehensive global network where data remains within the originating institutions,” he stated. This means that clinical insights gathered can contribute to evidence that withstands investigative scrutiny amidst the increasing integration of AI in research.

Chief Medical Informatics Officer Matvey Palchuk emphasizes the necessity of this validation, stating, “No system can extract every detail from clinical notes in both an accurate and meaningful fashion due to the inherent complexities of clinical language.” TriNetX's aim is to provide data that researchers can rely upon to generate trustworthy evidence.

Currently, the NLP-extracted Facts from Clinical Notes encompass a variety of fields, including social history (smoking and alcohol status), cardiology, pulmonology, general oncology, biomarkers, liver disease, and much more. These resources are readily accessible to members of the TriNetX network, adding substantial value to their research and clinical insights. For more information and to discover the vast potential of these advancements, visit TriNetX's clinical notes page.

About TriNetX, LLC

TriNetX is recognized as The Global Truth Engine for Better Human Health®. With a secure and interconnected network spanning over 14,200 clinical sites across more than 20 countries, TriNetX is committed to delivering credible evidence that meets the highest standards of scientific evaluation. By integrating universal data with unmatched integrity and a unique blend of AI and human intelligence, TriNetX aims to hasten discoveries and ultimately improve health outcomes worldwide. For more information, you can access their official website or follow them on LinkedIn.