Development of a Natural Language Processing Engine to Generate Bladder Cancer Pathology Data for Health Services Research

263 Background: Till this day urologists are waiting on symptomatic persons to initiate any diagnostic work-up to identify bladder cancer (BC) patients. In result we diagnose a quarter to a third of our patients as muscle-invasive cancers. The open-access questionnaire RiskCheck bladder cancer (RCBC) was proven in a pilot-study in daily routine from German urologists organized in the health services research foundation IQUO on asymptomatic patients. Methods: The open-access RCBC questionnaire was used in urological offices to check asymptomatic patients for their BC risk exposure (personal, smoking, occupation and medical induced). The tool delivers the classical risk stratification in low- intermediate- and high risk. All people with intermediate and high risk were checked for tumor presence by urine diagnostics and in case of suspect results controlled by cystoscopy. Statistical analysis was made by IBM-SPSS 19 for incidence distribution and correlation between risk stratification and tumor detection was proven by classification tree analysis, significance p < 0.05. Results: Out of 196 checked asymptomatic persons 185 (93.4%) were negative for tumor and 11 had a detectable tumor. In the group of NED 125 (68.1%) persons were classified as low risk, 26 (15.7%) as intermediate and 30 (16.2%) as high risk. Out of the 11 detected tumors 9 were at intermediate or high risk (81.8%). This resulted in an over all detection rate of 5.6% and focused on the risk population of 13.2%. The association of tumor presence and increased risk was significant (p < 0.01). Compared to the western incidence rates this is an increase in effectiveness of 377. Conclusions: Risk-adapted screening in bladder cancer delivers a reasonable approach to diagnose bladder cancer before emerging symptoms. The questionnaire RCBC integrates evidence based bladder cancer inductors, is easy in use and as a open-access tool available in 10 languages via the Internet ( www.riskcheck-bladder-cancer.info) .

Download Full-text

Identifying Bladder Cancer Stage And Use Of Chemotherapy In The Electronic Medical Record: How Reliable Is Natural Language Processing?

Proceedings of IMPRS ◽

10.18060/25780 ◽

2021 ◽

Vol 4 (1) ◽

Author(s):

Nirupama Devanathan ◽

David Haggstrom ◽

Clint Cary

Keyword(s):

Bladder Cancer ◽

Natural Language Processing ◽

Natural Language ◽

Language Processing ◽

Chart Review ◽

Clinical Stage ◽

Treatment Algorithm ◽

Clinical Staging ◽

Muscle Invasive ◽

Line Of Therapy

Background: Large automated electronic medical record (EMR) databases, together with natural language processing (NLP) algorithms, have the potential to be valuable tools in studying the patterns and effectiveness of treatment. Therefore, the current study sought to develop novel tools to identify bladder cancer cases, their clinical stage, and the chemotherapy they receive in electronic medical records. Methods: EMR data were obtained from Indiana University Health hospitals from 2008 to 2015. We developed 2 novel algorithms using natural language processing (NLP) on unstructured data to identify (a) bladder cancer cases and clinical stage, and (b) chemotherapy names and line of chemotherapy. The sensitivity, specificity, PPV, and NPV for the clinical staging and treatment algorithm were calculated against the gold standard of manual chart review Results: A total of 2,559 unique bladder cancer patients were identified and stratified using the clinical staging algorithm, defined as metastatic, muscle invasive, or non-muscle invasive. We identified 657 metastatic cases, 567 muscle invasive cases, and 604 non-muscle invasive cases. Further, we calculated the PPV for metastatic cases as 69.9%, muscle invasive as 80.4%, and non-muscle invasive as 79.1%. Next, the treatment algorithm was applied to metastatic patients to identify the type of chemotherapy received and 1st or 2nd line of therapy. The PPV for identifying the 1st and 2nd lines were 70.5% and 55.6%, respectively. The PPV for gemcitabine/carboplatin or cisplatin was 57.5%, but for methotrexate, vinblastine, doxorubicin, cisplatin, was 37.5%. Conclusion and Potential Impact: The performance of the algorithm demonstrates the potential for NLP to identify cancer cases, stage, and presence of treatment. While providing meaningful information, the accuracy of the approach suggests that a hybrid method using both NLP algorithms and manual chart review remains the most robust approach. The low performance of the algorithm to identify line of therapy further highlights the need for further NLP development in this area and emphasizes the ongoing need for either human entry or review of structured data.

Download Full-text