The life sciences industry is under constant pressure to enhance patient care, provide accurate patient outcomes, and measure results. The key issue faced by auditors during site audits is that most of the data is in the form of free text, which takes considerable time to read, evaluate, and assess the risk factors of each site. The life sciences organizations need to innovate and evolve the care paradigm while ensuring appropriate outcomes through proactive site audits.
Virtusa’s Site Monitoring Analytics and Anomaly Detection solution, powered by AWS Comprehend, analyses the reports and provides feedback on each comment with risk prediction and sentiment analysis for each site visit. The solution provides the resulting data to the auditors highlighting the critical risks for a quick site review and assessment to perform proactive audits.
Automated processes support the auditors during the verification of the Outbound Submission Tracking distribution rules, where the auditors must check the rules against the product lists, agreements, and legislation requirements. We have also developed an Auto Data Sampler where a stratified sample selected by country or product name, best represents the population by allotting specific weights to risk-related issues like Out of Compliance / Delayed Submission.
Virtusa’s site monitoring analytics and anomaly detection solution:
- Assesses the quality and compliance risk from inspection/audit findings
- Screens, predicts, and alerts on risks
- Extracts insights on site, patient, or study specific information not found in other sources
- Uses machine learning to mine historical data and convert text to risk scores, sentiment, AE issues, deviations from protocol or consent, and staff training
- Trains the AI model with the required categories, variables extracted from keywords using AWS Comprehend and random forest algorithms
With the site monitoring analytics and anomaly detection solution, our clients can:
- Reduce manual effort required to go through the long unstructured text and find similar records
- Rapidly detect potential risk or issues to enable proactive audits
- Classify the issues identified and received from data extraction
- Mine data from system logs as an additional source of data and to detect non-standard procedures and outliers
- Evaluate various topic modeling frameworks
- Get visualizations and dashboards to monitor the compliance level of a site quickly, observe trends, and graph by most common issues