Developing a Data and Analytics Platform to Enable a Breast Cancer Learning Health System at a Regional Cancer Center

Abstract

PURPOSE This study documents the creation of automated, longitudinal, and prospective data and analytics platform for breast cancer at a regional cancer center. This platform combines principles of data warehousing with natural language processing (NLP) to provide the integrated, timely, meaningful, high-quality, and actionable data required to establish a learning health system. METHODS Data from six hospital information systems and one external data source were integrated on a nightly basis by automated extract/transform/load jobs. Free-text clinical documentation was processed using a commercial NLP engine. RESULTS The platform contains 141 data elements of 7,019 patients with newly diagnosed breast cancer who received care at our regional cancer center from January 1, 2014, to June 3, 2022. Daily updating of the database takes an average of 56 minutes. Evaluation of the tuning of NLP jobs found overall high performance, with an F1 of 1.0 for 19 variables, with a further 16 variables with an F1 of > 0.95. CONCLUSION This study describes how data warehousing combined with NLP can be used to create a prospective data and analytics platform to enable a learning health system. Although upfront time investment required to create the platform was considerable, now that it has been developed, daily data processing is completed automatically in less than an hour.

Real-World Treatment Patterns and Clinical Outcomes among Patients Receiving CDK4/6 Inhibitors for Metastatic Breast Cancer in a Canadian Setting Using AI-Extracted Data

“…Further, the median (95% CI) for TTNT1 was 42.3 (41.2, NA) months for patients receiving a CDK4/6i in 1L, which is longer than the median rwPFS for palbociclib combination treatment from the US DeMichele et al ( 2022) study (20.0 months [95% CI, 17.5-21.9]) for 1L [11]. Additionally, the validation metrics for AI-extracted data are consistent with the previous validations of DARWEN TM , which has been evaluated against a manual abstraction for the same clinical features in breast cancer [25], lung cancer [18,[30][31][32][33][34], ambulatory care diseases [23], and dermatology [28] at multiple Canadian institutions.…”

“…Additionally, AI tools used for the extraction of clinical text can make sense of and analyze vast amounts of unstructured clinical text from pathology reports, clinical notes, and radiology reports. These tools, such as DARWEN TM , are being used for patient and disease identification, pharmacovigilance, and the development of learning health systems [23][24][25][26][27][28]. However, many tools, such as ClinicalBERT, rely on open-source datasets, such as the MIMIC-III dataset of de-identified hospital records from intensive care units [40,41].…”

Real-World Outcomes of Patients with Advanced Epidermal Growth Factor Receptor-Mutated Non-Small Cell Lung Cancer in Canada Using Data Extracted by Large Language Model-Based Artificial Intelligence

“…This was a retrospective cohort study of data elements from EHRs stored at the UHN-PMCC using AI technology. The AI engine combines large language models and an ensemble of other techniques that have previously been evaluated and validated against manual abstraction across multiple disease domains, including lung cancer [22,25], breast cancer [26], dermatology [27], and infectious diseases [28] at multiple Canadian institutions, including the UHN-PMCC.…”

Health at a Glance 2023

“…Improved health literacy has been shown to improve trust among the public regarding health communications (Paige, Krieger and Stellefson, 2016 [33]). Hence, actions to address both digital literacy and health literacy are important parts of digital health readiness.…”