Open Source Infrastructure for Health Care Data Integration and Machine Learning Analyses

Abstract

PURPOSE We have created a cloud-based machine learning system (CLOBNET) that is an open-source, lean infrastructure for electronic health record (EHR) data integration and is capable of extract, transform, and load (ETL) processing. CLOBNET enables comprehensive analysis and visualization of structured EHR data. We demonstrate the utility of CLOBNET by predicting primary therapy outcomes of patients with high-grade serous ovarian cancer (HGSOC) on the basis of EHR data. MATERIALS AND METHODS CLOBNET is built using open-source software to make data preprocessing, analysis, and model training user friendly. The source code of CLOBNET is available in GitHub. The HGSOC data set was based on a prospective cohort of 208 patients with HGSOC who were treated at Turku University Hospital, Finland, from 2009 to 2019 for whom comprehensive clinical and EHR data were available. RESULTS We trained machine learning (ML) models using clinical data, including a herein developed dissemination score that quantifies the disease burden at the time of diagnosis, to identify patients with progressive disease (PD) or a complete response (CR) on the basis of RECIST (version 1.1). The best performance was achieved with a logistic regression model, which resulted in an area under receiver operating characteristic curve (AUROC) of 0.86, with a specificity of 73% and a sensitivity of 89%, when it classified between patients who experienced PD and CR. CONCLUSION We have developed an open-source computational infrastructure, CLOBNET, that enables effective and rapid analysis of EHR and other clinical data. Our results demonstrate that CLOBNET allows predictions to be made on the basis of EHR data to address clinically relevant questions.

A longitudinal analysis of CA125 glycoforms in the monitoring and follow up of high grade serous ovarian cancer

“…The operating team systematically assessed and documented the disease spread and tumor volume in the peritoneal cavity and retroperitoneum with a standardized 16-part questionnaire during surgery. Each abdominal site and possible metastasis was included, and a recently validated disease dissemination score ranging from 0 to 21 was calculated ( Table 2) [28]. We divided patients to a low tumor load group (score 0-11) and a high tumor load group (score 12-21) based on the disease dissemination score.…”

HE4 in the evaluation of tumor load and prognostic stratification of high grade serous ovarian carcinoma

“…The operating gynecologic oncologist assessed the disease stage as stated by the International Federation of Gynecology and Obstetrics (FIGO) 2014 classification and a pathologist confirmed it from tissue biopsies. The operating team recorded the disease spread in the abdominal cavity and retroperitoneum to calculate a previously described disease dissemination score (range 0-21) (Table S1) [24]. The disease dissemination score showed prognostic value in a previous study [24].…”

“…The operating team recorded the disease spread in the abdominal cavity and retroperitoneum to calculate a previously described disease dissemination score (range 0-21) (Table S1) [24]. The disease dissemination score showed prognostic value in a previous study [24]. Patients were divided into a low tumor load group (dissemination score 0-12) and a high tumor load group (dissemination score [13][14][15][16][17][18][19][20][21].…”

A Review of AI and Data Science Support for Cancer Management

“…The data integration approaches from recent years use a combination of both approaches. Several papers [46][47][48][58][59][60] present ETL database functions that allow pulling data from a source database and placing it into a target database. The first three of these papers define manual processes by which local sites define mappings between source data elements (SDE) to a set of needed common data elements (CDE).…”

“…This clustering is relevant to discriminate between different breast cancer subtypes and to identify their relations and has the strength that it can be performed in an unsupervised manner, removing much of the burden of integration based on information models and ontologies. Isoviita et al [48] present an open-source, cloud-based machine learning system where datasets from multiple (live) sources (EHR databases and research databases) are integrated using extract, transform, load (ETL) processes and melded into a single database, but with minimal transformations. These merged but heterogeneous data are used for the training of ML predictive models.…”