KEY TAKEAWAYS
- The study aimed to develop an automatic rule-based algorithm for detecting OC recurrence using minimally preprocessed EHR data.
- The study concluded that the rule-based algorithm accurately identified initial ovarian cancer recurrence, supporting enhanced large-scale EHR analysis for clinical research.
Given the lack of explicit documentation of ovarian cancer recurrence in electronic health records (EHRs), significant manual chart review is needed.
Sanghee Lee and the team aimed to create an automated rule-based algorithm for identifying recurrent ovarian cancer using minimally processed EHR data.
The Auto-Recur algorithm, employing data from imaging reports (PET-CT, CT, MRI), biomarker (CA125), and treatment records (surgery, chemotherapy, radiotherapy), was designed to detect initial ovarian cancer recurrence. It comprises three standalone algorithms (images, biomarkers, treatments) and hybrid algorithms (combinations of these). Performance evaluation included sensitivity, specificity, and accuracy of recurrence time detection, with comparison to retrospective chart review for estimating recurrence-free survival probabilities.
The results showed that the Auto-Recur system significantly decreased resource and time requirements, saving around 1,340 days for 100,000 patients compared to conventional chart review. Among the algorithms, the hybrid variant, combining imaging, biomarker, and treatment data, demonstrated the highest efficiency (sensitivity: 93.4%, specificity: 97.4%) and precise recurrence time capture (average time error: 8.5 days). The estimated 3-year recurrence-free survival probability (44%) closely matched that from retrospective chart review (45%, log-rank P value = .894).
The study concluded that the rule-based algorithm successfully identified initial ovarian cancer recurrence in extensive EHR data, aligning closely with recurrence-free survival estimates from traditional retrospective chart reviews. These findings promote broader EHR analysis, fostering opportunities for clinical research enhancement.
The funding source did not contribute to the study design, data collection, analysis, or interpretation.
Source: https://pubmed.ncbi.nlm.nih.gov/38442323/
Lee S, Kim JH, Ha HI, et al. (2024). “Development of an Automatic Rule-Based Algorithm for the Detection of Ovarian Cancer Recurrence From Electronic Health Records.” JCO Clin Cancer Inform. 2024 Mar;8:e2300150. doi: 10.1200/CCI.23.00150. PMID: 38442323.
