Leveraging Genetic Reports and Electronic Health Records for the Prediction of Primary Cancers: Algorithm Development and Validation Study

Abstract

Background Precision oncology has the potential to leverage clinical and genomic data in advancing disease prevention, diagnosis, and treatment. A key research area focuses on the early detection of primary cancers and potential prediction of cancers of unknown primary in order to facilitate optimal treatment decisions. Objective This study presents a methodology to harmonize phenotypic and genetic data features to classify primary cancer types and predict cancers of unknown primaries. Methods We extracted genetic data elements from oncology genetic reports of 1011 patients with cancer and their corresponding phenotypical data from Mayo Clinic’s electronic health records. We modeled both genetic and electronic health record data with HL7 Fast Healthcare Interoperability Resources. The semantic web Resource Description Framework was employed to generate the network-based data representation (ie, patient-phenotypic-genetic network). Based on the Resource Description Framework data graph, Node2vec graph-embedding algorithm was applied to generate features. Multiple machine learning and deep learning backbone models were compared for cancer prediction performance. Results With 6 machine learning tasks designed in the experiment, we demonstrated the proposed method achieved favorable results in classifying primary cancer types (area under the receiver operating characteristic curve [AUROC] 96.56% for all 9 cancer predictions on average based on the cross-validation) and predicting unknown primaries (AUROC 80.77% for all 8 cancer predictions on average for real-patient validation). To demonstrate the interpretability, 17 phenotypic and genetic features that contributed the most to the prediction of each cancer were identified and validated based on a literature review. Conclusions Accurate prediction of cancer types can be achieved with existing electronic health record data with satisfactory precision. The integration of genetic reports improves prediction, illustrating the translational values of incorporating genetic tests early at the diagnosis stage for patients with cancer.

Artificial Intelligence Models in Health Information Exchange: A Systematic Review of Clinical Implications

“…These AI models and algorithms may accomplish various tasks, such as data extraction, clinical decision assistance, and prognosis prediction. In addition, AI may forecast multiple health-related results, such as cancer, sepsis, heart failure, in-hospital cardiac arrest, and COVID-19-related resource utilization [ 30 , 32 , 40 , 41 , 42 , 43 ]. Several measures, including area under the curve (AUC), precision score, positive predictive value (PPV), negative predictive value (NPV), sensitivity, specificity, calibration, and F-measure, were used to evaluate the performance of algorithms.…”

“…FHIR is the most current HL7 system standard [ 54 ]. It was first introduced in March 2014, and multiple technical design studies conducted between 2018 and 2022 favored FHIR as their preferred standard [ 40 , 41 , 42 , 43 ]. HL7 messaging systems were used by [ 31 , 32 , 37 , 38 ] to gather their input information, and some authors tried to improve their data collection quality using HL7 version 2.…”

“…NLP, DNN, and ML models can categorize primary cancer types and forecast cancers of unclear origin [ 40 ]. NLP may be the best option for recognizing clinical notes that are not coded.…”

Applying Natural Language Processing to Textual Data From Clinical Data Warehouses: Systematic Review

“…Numerous studies detail the application of the Medical Language Extraction and Encoding System (MedLEE) for clinical concepts [24,[27][28][29][32][33][34][35][36]50,51,121] or medication [103,104,111] extraction, as well as UMLS coding. The extraction and mapping of clinical information from clinical notes to UMLS has also been accomplished using the clinical Text Analysis and Knowledge Extraction System (cTAKES) [16,17,20,22,100,129,134,168], MetaMap [31,37,38,47], MedTagger [44,45,67,78,86,105], and the National Center for Biomedical Ontology (NCBO) Annotator [97,99,106,107,109,114]. Extracted concepts can be mapped to other standard ontologies and terminologies, such as SNOMED-CT [27].…”

FHIR-Ontop-OMOP: Building clinical knowledge graphs in FHIR RDF with the OMOP Common data Model

“…The combination of FHIR, KGs and the Semantic Web enables a new paradigm to build explainable AI applications in healthcare. A few of such FHIR-based applications are emerging, including 1) a KG generation tool known as NLP2FHIR developed for standardizing and integrating unstructured and structured EHR data in FHIR [ 18 ]; 2) a FHIR-based EHR phenotyping framework using machine learning and deep learning techniques developed for effective data integration and accurate phenotyping [ 19 ]; and 3) FHIR RDF data is used to build AI algorithms to predict primary cancers, showing accurate prediction of cancer types can be achieved with existing EHR data and genetic report data [ 20 ].…”