Recommendations for patient similarity classes: results of the AMIA 2019 workshop on defining patient similarity

A large-scale dataset of patient summaries for retrieval-based clinical decision support systems

Retrieval-based Clinical Decision Support (ReCDS) can aid clinical workflow by providing relevant literature and similar patients for a given patient. However, the development of ReCDS systems has been severely obstructed by the lack of diverse patient collections and publicly available large-scale patient-level annotation datasets. In this paper, we collect a novel dataset of patient summaries and relations called PMC-Patients to benchmark two ReCDS tasks: Patient-to-Article Retrieval (ReCDS-PAR) and Patient-to-Patient Retrieval (ReCDS-PPR). Specifically, we extract patient summaries from PubMed Central articles using simple heuristics and utilize the PubMed citation graph to define patient-article relevance and patient-patient similarity. PMC-Patients contains 167k patient summaries with 3.1 M patient-article relevance annotations and 293k patient-patient similarity annotations, which is the largest-scale resource for ReCDS and also one of the largest patient collections. Human evaluation and analysis show that PMC-Patients is a diverse dataset with high-quality annotations. We also implement and evaluate several ReCDS systems on the PMC-Patients benchmarks to show its challenges and conduct several case studies to show the clinical utility of PMC-Patients.

Gene partners of the EWSR1 fusion may represent molecularly distinct entities

EWSR1 fusions are highly promiscuous and are associated with unique malignancies, clinical phenotypes, and molecular subtypes. However, rare fusion partners (RFP) of EWSR1 has not been well described. Here, we conducted a cross-sectional, retrospective study of 1,140 unique tumors harboring EWSR1 fusions. We identified 64 unique fusion partners. RFPs were identified more often in adults than children. Alterations in cell cycle control and DNA damage response genes as driving the differences between fusion partners. Potentially clinically actionable genomic variants were more prevalent in tumors harboring RFP than common fusions. While the data presented here is limited, tumors harboring RFP of EWSR1 may represent molecularly distinct entities and may benefit from further molecular testing to identify targeted therapeutic options.

Comparison of variants in TPMT and NUDT15 between sequencing and genotyping methods in a multistate pediatric institution

The risk of severe adverse events related to thiopurine therapy can be reduced by personalizing dosing based on TPMT and NUDT15 genetic polymorphisms. However, the optimal genetic testing platform has not yet been established. In this study, we report on the TPMT and NUDT15 genotypes and phenotypes generated from 320 patients from a multicenter pediatric healthcare system using both Sanger sequencing and polymerase chain reaction genotyping (hereafter: genotyping) methods to determine the appropriateness of genotyping in our patient population. Sanger sequencing identified variant TPMT alleles including *3A (8, 3.2% of alleles), *3C (4, 1.6%), and *2 (1, 0.4%), and NUDT15 alleles including *2 (5, 3.6%) and *3 (1, 0.7%). For genotyped patients, variants identified in TPMT included *3A (12, 3.1%), *3C (4, 1%), *2 (2, 0.5%), and *8 (1, 0.25%), whereas NUDT15 included *4 (2, 1.9%) and *2 or *3 (1, 1%). Between Sanger sequencing and genotyping, no significant difference in allele, genotype, or phenotype frequency was identified for either TPMT or NUDT15. All patients who were tested using Sanger sequencing would have been accurately phenotyped for either TPMT (124/124), NUDT15 (69/69), or both genes (68/68) if they were assayed using the genotyping method. Considering 193 total TPMT and NUDT15 Sanger Sequencing tests reviewed, all tests would have resulted in an appropriate clinical recommendation if the test had instead been conducted using the comparison genotyping platforms. These results suggest that, in this study population, genotyping would be sufficient to provide accurate phenotype calls and clinical recommendations.

Integrating pharmacogenomic testing into paired germline and somatic genomic testing in patients with cancer

Precision medicine has revolutionized clinical care for patients with cancer through the development of targeted therapy, identification of inherited cancer predisposition syndromes and the use of pharmacogenetics to optimize pharmacotherapy for anticancer drugs and supportive care medications. While germline (patient) and somatic (tumor) genomic testing have evolved separately, recent interest in paired germline/somatic testing has led to an increase in integrated genomic testing workflows. However, paired germline/somatic testing has generally lacked the incorporation of germline pharmacogenomics. Integrating pharmacogenomics into paired germline/somatic genomic testing would be an efficient method for increasing access to pharmacogenomic testing. In this perspective, the authors argue for the benefits of implementing a comprehensive approach integrating somatic and germline testing that is inclusive of pharmacogenomics in clinical practice.

Using similar patients to predict complication in patients with diabetes, hypertension, and lipid disorder: a domain knowledge-infused convolutional neural network approach

OBJECTIVE

This study aims to develop a convolutional neural network-based learning framework called domain knowledge-infused convolutional neural network (DK-CNN) for retrieving clinically similar patient and to personalize the prediction of macrovascular complication using the retrieved patients.

MATERIALS AND METHODS

We use the electronic health records of 169 434 patients with diabetes, hypertension, and/or lipid disorder. Patients are partitioned into 7 subcohorts based on their comorbidities. DK-CNN integrates both domain knowledge and disease trajectory of patients over multiple visits to retrieve similar patients. We use normalized discounted cumulative gain (nDCG) and macrovascular complication prediction performance to evaluate the effectiveness of DK-CNN compared to state-of-the-art models. Ablation studies are conducted to compare DK-CNN with reduced models that do not use domain knowledge as well as models that do not consider short-term, medium-term, and long-term trajectory over multiple visits.

RESULTS

Key findings from this study are: (1) DK-CNN is able to retrieve clinically similar patients and achieves the highest nDCG values in all 7 subcohorts; (2) DK-CNN outperforms other state-of-the-art approaches in terms of complication prediction performance in all 7 subcohorts; and (3) the ablation studies show that the full model achieves the highest nDCG compared with other 2 reduced models.

DISCUSSION AND CONCLUSIONS

DK-CNN is a deep learning-based approach which incorporates domain knowledge and patient trajectory data to retrieve clinically similar patients. It can be used to assist physicians who may refer to the outcomes and past treatments of similar patients as a guide for choosing an effective treatment for patients.

Deep Dynamic Patient Similarity Analysis: Model Development and Validation in ICU

BACKGROUND

Personalized medicine requires the patient similarity analysis for providing specific treatments tailed for each patient. However, the patient similarity analysis in personalized clinical scenarios encounters challenges, which are twofold. First, heterogeneous and multi-type data are usually recorded to Electronic Health Records (EHRs) during the course of admissions, which makes it difficult to measure the patient similarity. Second, disease progression manifests diverse disease states at different times, which brings sequential complexity to dynamically retrieve similar patients' sequences.

MATERIALS AND METHODS

To overcome the above-mentioned challenges, we propose a novel dynamic patient similarity analysis model based on deep learning. Specifically, the proposed model embeds the multi-type and heterogeneous data into hidden representations with a specially designed embedding and attention module. Thereafter, the proposed model retrieves similar patients' sequences based on these hidden representations in a dynamic manner. More importantly, we adopt two clinical tasks, i.e., diagnosis prediction and medication recommendation, to validate the effectiveness of the proposed model. It is worth noticing that the proposed model integrates a drug-drug interaction (DDI) knowledge graph in the medication recommendation task to reduce adverse reactions caused by combinational treatments, such that a more rational strategy can be realized. We evaluate our proposed model using the critical care database MIMIC-III, which includes 5,430 patients covering 14,096 clinical visits.

RESULTS

The proposed model outperforms several state-of-the-art methods. For diagnosis prediction, the average PR-AUC score of the proposed model reaches 0.6200, which is significantly higher than that of the baseline models (0.2497∼0.5407). Meanwhile, for medication recommendation, the average PR-AUC of the proposed model is 0.6682 (Jaccard: 0.4070; F1: 0.5672; Recall: 0.7832) whereas the K-nearest model can only reach 0.3805 (Jaccard: 0.3911; F1: 0.5465; Recall: 0.5705). In addition, our proposed model achieves a lower DDI rate.

CONCLUSION

We propose a novel dynamic patient similarity analysis model, which can be implemented into a decision support system for clinical tasks including diagnosis prediction, surgical procedure selection, medication recommendation, etc. Also, the proposed model serves as an explainable protocol in clinical practice thanks to its analogy to real clinical reasoning where a doctor diagnoses diseases and prescribes medications according to the previous cured patients empirically.

El-Kassabi H, Serhani MA, Oulhaj A, Khalil K. A Novel Patient Similarity Network (PSN) Framework Based on Multi-Model Deep Learning for Precision Medicine

Precision medicine can be defined as the comparison of a new patient with existing patients that have similar characteristics and can be referred to as patient similarity. Several deep learning models have been used to build and apply patient similarity networks (PSNs). However, the challenges related to data heterogeneity and dimensionality make it difficult to use a single model to reduce data dimensionality and capture the features of diverse data types. In this paper, we propose a multi-model PSN that considers heterogeneous static and dynamic data. The combination of deep learning models and PSN allows ample clinical evidence and information extraction against which similar patients can be compared. We use the bidirectional encoder representations from transformers (BERT) to analyze the contextual data and generate word embedding, where semantic features are captured using a convolutional neural network (CNN). Dynamic data are analyzed using a long-short-term-memory (LSTM)-based autoencoder, which reduces data dimensionality and preserves the temporal features of the data. We propose a data fusion approach combining temporal and clinical narrative data to estimate patient similarity. The experiments we conducted proved that our model provides a higher classification accuracy in determining various patient health outcomes when compared with other traditional classification algorithms.

Drivers of genomic loss of heterozygosity in leiomyosarcoma are distinct from carcinomas

Leiomyosarcoma (LMS) is a rare, aggressive, mesenchymal tumor. Subsets of LMS have been identified to harbor genomic alterations associated with homologous recombination deficiency (HRD); particularly alterations in BRCA2. Whereas genomic loss of heterozygosity (gLOH) has been used as a surrogate marker of HRD in other solid tumors, the prognostic or clinical value of gLOH in LMS (gLOH-LMS) remains poorly defined. We explore the genomic drivers associated with gLOH-LMS and their clinical import. Although the distribution of gLOH-LMS scores are similar to that of carcinomas, outside of BRCA2, there was no overlap with previously published gLOH-associated genes from studies in carcinomas. We note that early stage tumors with elevated gLOH demonstrated a longer disease-free interval following resection in LMS patients. Taken together, and despite similarities to carcinomas in gLOH distribution and clinical import, gLOH-LMS are driven by different genomic signals. Additional studies will be required to isolate and confirm the unique differences in biological factors driving these differences.

Can Machine Learning from Real-World Data Support Drug Treatment Decisions? A Prediction Modeling Case for Direct Oral Anticoagulants

BACKGROUND

Decision making for the "best" treatment is particularly challenging in situations in which individual patient response to drugs can largely differ from average treatment effects. By estimating individual treatment effects (ITEs), we aimed to demonstrate how strokes, major bleeding events, and a composite of both could be reduced by model-assisted recommendations for a particular direct oral anticoagulant (DOAC).

METHODS

In German claims data for the calendar years 2014-2018, we selected 29 901 new users of the DOACs rivaroxaban and apixaban. Random forests considered binary events within 1 y to estimate ITEs under each DOAC according to the X-learner algorithm with 29 potential effect modifiers; treatment recommendations were based on these estimated ITEs. Model performance was evaluated by the c-for-benefit statistics, absolute risk reduction (ARR), and absolute risk difference (ARD) by trial emulation.

RESULTS

A significant proportion of patients would be recommended a different treatment option than they actually received. The stroke model significantly discriminated patients for higher benefit and thus indicated improved decisions by reduced outcomes (c-for-benefit: 0.56; 95% confidence interval [0.52; 0.60]). In the group with apixaban recommendation, the model also improved the composite endpoint (ARR: 1.69 % [0.39; 2.97]). In trial emulations, model-assisted recommendations significantly reduced the composite event rate (ARD: -0.78 % [-1.40; -0.03]).

CONCLUSIONS

If prescribers are undecided about the potential benefits of different treatment options, ITEs can support decision making, especially if evidence is inconclusive, risk-benefit profiles of therapeutic alternatives differ significantly, and the patients' complexity deviates from "typical" study populations. In the exemplary case for DOACs and potentially in other situations, the significant impact could also become practically relevant if recommendations were available in an automated way as part of decision making.HighlightsIt was possible to calculate individual treatment effects (ITEs) from routine claims data for rivaroxaban and apixaban, and the characteristics between the groups with recommendation for one or the other option differed significantly.ITEs resulted in recommendations that were significantly superior to usual (observed) treatment allocations in terms of absolute risk reduction, both separately for stroke and in the composite endpoint of stroke and major bleeding.When similar patients from routine data were selected (precision cohorts) for patients with a strong recommendation for one option or the other, those similar patients under the respective recommendation showed a significantly better prognosis compared with the alternative option.Many steps may still be needed on the way to clinical practice, but the principle of decision support developed from routine data may point the way toward future decision-making processes.

Using Domain Knowledge and Data-Driven Insights for Patient Similarity Analytics

Patient similarity analytics has emerged as an essential tool to identify cohorts of patients who have similar clinical characteristics to some specific patient of interest. In this study, we propose a patient similarity measure called D3K that incorporates domain knowledge and data-driven insights. Using the electronic health records (EHRs) of 169,434 patients with either diabetes, hypertension or dyslipidaemia (DHL), we construct patient feature vectors containing demographics, vital signs, laboratory test results, and prescribed medications. We discretize the variables of interest into various bins based on domain knowledge and make the patient similarity computation to be aligned with clinical guidelines. Key findings from this study are: (1) D3K outperforms baseline approaches in all seven sub-cohorts; (2) our domain knowledge-based binning strategy outperformed the traditional percentile-based binning in all seven sub-cohorts; (3) there is substantial agreement between D3K and physicians (κ = 0.746), indicating that D3K can be applied to facilitate shared decision making. This is the first study to use patient similarity analytics on a cardiometabolic syndrome-related dataset sourced from medical institutions in Singapore. We consider patient similarity among patient cohorts with the same medical conditions to develop localized models for personalized decision support to improve the outcomes of a target patient.

Multiscale-omic assessment of EWSR1-NFATc2 fusion positive sarcomas identifies the mTOR pathway as a potential therapeutic target

Sarcomas harboring EWSR1-NFATc2 fusions have historically been categorized and treated as Ewing sarcoma. Emerging evidence suggests unique molecular characteristics and chemotherapy sensitivities in EWSR1-NFATc2 fusion positive sarcomas. Comprehensive genomic profiles of 1024 EWSR1 fusion positive sarcomas, including 14 EWSR1-NFATc2 fusions, were identified in the FoundationCore® database. Additional data from the Gene Expression Omnibus, the Genomics of Drug Sensitivity in Cancer and The Cancer Genome Atlas datasets were included for analysis. EWSR1-NFATc2 fusion positive sarcomas were genomically distinct from traditional Ewing sarcoma and demonstrated upregulation of the mTOR pathway. We also present a case of a 58-year-old male patient with metastatic EWSR1-NFATc2 fusion positive sarcoma who achieved 47 months of disease stabilization when treated with combination mTOR and VEGF inhibition. EWSR1-NFATc2 fusion positive sarcomas are molecularly distinct entities with overactive mTOR signaling; which may be therapeutically targetable. These findings support the use of precision medicine in the Ewing family of tumors.

OSIRIS: A Minimum Data Set for Data Sharing and Interoperability in Oncology

PURPOSE

Many institutions throughout the world have launched precision medicine initiatives in oncology, and a large amount of clinical and genomic data is being produced. Although there have been attempts at data sharing with the community, initiatives are still limited. In this context, a French task force composed of Integrated Cancer Research Sites (SIRICs), comprehensive cancer centers from the Unicancer network (one of Europe's largest cancer research organization), and university hospitals launched an initiative to improve and accelerate retrospective and prospective clinical and genomic data sharing in oncology.

MATERIALS AND METHODS

For 5 years, the OSIRIS group has worked on structuring data and identifying technical solutions for collecting and sharing them. The group used a multidisciplinary approach that included weekly scientific and technical meetings over several months to foster a national consensus on a minimal data set.

RESULTS

The resulting OSIRIS set and event-based data model, which is able to capture the disease course, was built with 67 clinical and 65 omics items. The group made it compatible with the HL7 Fast Healthcare Interoperability Resources (FHIR) format to maximize interoperability. The OSIRIS set was reviewed, approved by a National Plan Strategic Committee, and freely released to the community. A proof-of-concept study was carried out to put the OSIRIS set and Common Data Model into practice using a cohort of 300 patients.

CONCLUSION

Using a national and bottom-up approach, the OSIRIS group has defined a model including a minimal set of clinical and genomic data that can be used to accelerate data sharing produced in oncology. The model relies on clear and formally defined terminologies and, as such, may also benefit the larger international community.