Network-based analysis of diagnosis progression patterns using claims data

Network Analysis and Machine Learning for Signal Detection and Prioritization Using Electronic Healthcare Records and Administrative Databases: A Proof of Concept in Drug-Induced Acute Myocardial Infarction

BACKGROUND

Safety signals for potential drug-induced adverse events (AEs) typically emerge from multiple data sources, primarily spontaneous reporting systems, despite known limitations. Increasingly, real-world data from sources such as electronic health records (EHRs) and administrative databases are leveraged for signal detection. Although network analysis has shown promise in mapping relationships between clinical attributes for signal detection in spontaneous reporting system databases, its application in real-world data from EHRs and administrative databases remains limited.

OBJECTIVE

This study aimed to evaluate the performance of network analysis in detecting safety signals within Italian administrative databases, using drug-induced acute myocardial infarction (AMI) as a proof of concept.

METHODS

We employed a case-crossover design to explore the association between drug exposure and AMI using the Healthcare Administrative Database of Mantova, Italy, from 2014 to 2018. Patients with their first AMI hospitalization were identified after a 365-day washout period to exclude prior hospitalizations. We constructed a network to analyse the relationships between prescribed drugs and diagnoses, represented as nodes, with undirected edges illustrating their interactions. For each patient with AMI, we identified all diagnoses and drugs recorded or redeemed within 365 days of the first AMI episode and generated various drug-diagnosis, drug-drug, and diagnosis-diagnosis pairs. We calculated the frequency of these pairs, and three types of edge weights quantified the strength of connections. We identified outlier drug-AMI pairs using a predictive score (F) based on frequency (C) and full edge weights (WF), with validation for known AMI associations. We prioritized signals using the F score, C of AMI, and WF, analysed through k-means clustering to identify patterns in the data.

RESULTS

From 2014 to 2018, a total of 3918 patients had an AMI, with 4686 AMI diagnoses. Of those, 2866 had prescriptions in the previous year, totalling 498,591 prescriptions. A network analysis identified 2968 unique nodes, revealing 529,935 diagnosis-diagnosis connections, 235,380 drug-diagnosis connections, and 102,831 drug-drug connections. The median number of connections (C) was 404 (Q1-Q3: 194-671) for drug nodes and 380 (Q1-Q3: 216-664) for diagnosis nodes. The median WF was 11.8 (Q1-Q3: 9-14), and the median F score across pairs was 0.1 (Q1-Q3: 0.1-0.3). A total of 249 potential safety signals were detected, with 63.4% aligning with known AEs. Among the remaining signals, 80 were prioritized, and five emerged as the highest priority: terazosin, tamsulosin, allopurinol, esomeprazole, and omeprazole.

CONCLUSIONS

Overall, our novel method demonstrates that network analysis is a valuable tool for signal detection and prioritization in drug-induced AEs based on EHRs and administrative databases.

Comorbidity Networks From Population-Wide Health Data: Aggregated Data of 8.9M Hospital Patients (1997-2014)

Comorbidity networks have become a valuable tool to support data-driven biomedical research. Yet, studies often are severely hindered by the availability of the necessary comprehensive data, often due to the sensitivity of health care information. This study presents a population-wide comorbidity network dataset derived from 45 million hospital stays of 8.9 million patients over 17 years in Austria. We present co-occurrence networks of hospital diagnoses, stratified by age, sex, and observation period in a total of 96 different subgroups. For each of these groups we report a range of association measures (e.g., count data, and odds ratios) for all pairs of diagnoses. The dataset provides the possibility to researchers to create their own, tailor-made comorbidity networks from real patient data that can be used as a starting point in quantitative and machine learning methods. This data platform is intended to lead to deeper insights into a wide range of epidemiological, public health, and biomedical research questions.

A network-based approach to explore comorbidity patterns among community-dwelling older adults living alone

The detailed comorbidity patterns of community-dwelling older adults have not yet been explored. This study employed a network-based approach to investigate the comorbidity patterns of community-dwelling older adults living alone. The sample comprised a cross-sectional cohort of adults 65 or older living alone in a Korean city (n = 1041; mean age = 77.7 years, 77.6% women). A comorbidity network analysis that estimates networks aggregated from measures of significant co-occurrence between pairs of diseases was employed to investigate comorbid associations between 31 chronic conditions. A cluster detection algorithm was employed to identify specific clusters of comorbidities. The association strength was expressed as the observed-to-expected ratio (OER). As a result, fifteen diseases were interconnected within the network (OER > 1, p-value < .05). While hypertension had a high prevalence, osteoporosis was the most central disease, co-occurring with numerous other diseases. The strongest associations among comorbidities were found between thyroid disease and urinary incontinence, chronic otitis media and osteoporosis, gastric duodenal ulcer/gastritis and anemia, and depression and gastric duodenal ulcer/gastritis (OER > 1.85). Three distinct clusters were identified as follows: (a) cataracts, osteoporosis, chronic otitis media, osteoarthritis/rheumatism, low back pain/sciatica, urinary incontinence, post-accident sequelae, and thyroid diseases; (b) hyperlipidemia, diabetes mellitus, and hypertension; and (c) depression, skin disease, gastric duodenal ulcer/gastritis, and anemia. The results may prove valuable in guiding the early diagnosis, management, and treatment of comorbidities in older adults living alone.

Comorbidity Patterns in Older Patients Undergoing Hip Fracture Surgery: A Comorbidity Network Analysis Study

Comorbidity network analysis (CNA) is a technique in which mathematical graphs encode correlations (edges) among diseases (nodes) inferred from the disease co-occurrence data of a patient group. The present study applied this network-based approach to identifying comorbidity patterns in older patients undergoing hip fracture surgery. This was a retrospective observational cohort study using electronic health records (EHR). EHR data were extracted from the one University Health System in the southeast United States. The cohort included patients aged 65 and above who had a first-time low-energy traumatic hip fracture treated surgically between October 1, 2015 and December 31, 2018 (n = 1,171). Comorbidity includes 17 diagnoses classified by the Charlson Comorbidity Index. The CNA investigated the comorbid associations among 17 diagnoses. The association strength was quantified using the observed-to-expected ratio (OER). Several network centrality measures were used to examine the importance of nodes, namely degree, strength, closeness, and betweenness centrality. A cluster detection algorithm was employed to determine specific clusters of comorbidities. Twelve diseases were significantly interconnected in the network (OER > 1, p-value < .05). The most robust associations were between metastatic carcinoma and mild liver disease, myocardial infarction and congestive heart failure, and hemi/paraplegia and cerebrovascular disease (OER > 2.5). Cerebrovascular disease, congestive heart failure, and myocardial infarction were identified as the central diseases that co-occurred with numerous other diseases. Two distinct clusters were noted, and the largest cluster comprised 10 diseases, primarily encompassing cardiometabolic and cognitive disorders. The results highlight specific patient comorbidities that could be used to guide clinical assessment, management, and targeted interventions that improve hip fracture outcomes in this patient group.

Conceptualised psycho-medical footprint for health status outcomes and the potential impacts for early detection and prevention of chronic diseases in the context of 3P medicine

Background

The Suboptimal Health Status Questionnaire-25 (SHSQ-25) is a distinctive medical psychometric diagnostic tool designed for the early detection of chronic diseases. However, the synaptic connections between the 25 symptomatic items and their relevance in supporting the monitoring of suboptimal health outcomes, which are precursors for chronic diseases, have not been thoroughly evaluated within the framework of predictive, preventive, and personalised medicine (PPPM/3PM). This baseline study explores the internal structure of the SHSQ-25 and demonstrates its discriminatory power to predict optimal and suboptimal health status (SHS) and develop photogenic representations of their distinct relationship patterns.

Methods

The cross-sectional study involved healthy Ghanaian participants (n = 217; aged 30-80 years; ~ 61% female), who responded to the SHSQ-25. The median SHS score was used to categorise the population into optimal and SHS. Graphical LASSO model and multi-dimensional scaling configuration methods were employed to describe the network structures for the two populations.

Results

We observed differences in the structural, node placement and node distance of the synaptic networks for the optimal and suboptimal populations. A statistically significant variance in connectivity levels was noted between the optimal (58 non-zero edges) and suboptimal (43 non-zero edges) networks (p = 0.024). Fatigue emerged as a prominently central subclinical condition within the suboptimal population, whilst the cardiovascular system domain had the greatest relevance for the optimal population. The contrast in connectivity levels and the divergent prominence of specific subclinical conditions across domain networks shed light on potential health distinctions.

Conclusions

We have demonstrated the feasibility of creating dynamic visualizers of the evolutionary trends in the relationships between the domains of SHSQ-25 relative to health status outcomes. This will provide in-depth comprehension of the conceptual model to inform personalised strategies to circumvent SHS. Additionally, the findings have implications for both health care and disease prevention because at-risk individuals can be predicted and prioritised for monitoring, and targeted intervention can begin before their symptoms reach an irreversible stage.

Supplementary information

The online version contains supplementary material available at 10.1007/s13167-023-00344-2.

Revealing the dynamic landscape of drug-drug interactions through network analysis

Introduction: The landscape of drug-drug interactions (DDIs) has evolved significantly over the past 60 years, necessitating a retrospective analysis to identify research trends and under-explored areas. While methodologies like bibliometric analysis provide valuable quantitative perspectives on DDI research, they have not successfully delineated the complex interrelations between drugs. Understanding these intricate relationships is essential for deciphering the evolving architecture and progressive transformation of DDI research structures over time. We utilize network analysis to unearth the multifaceted relationships between drugs, offering a richer, more nuanced comprehension of shifts in research focus within the DDI landscape. Methods: This groundbreaking investigation employs natural language processing, techniques, specifically Named Entity Recognition (NER) via ScispaCy, and the information extraction model, SciFive, to extract pharmacokinetic (PK) and pharmacodynamic (PD) DDI evidence from PubMed articles spanning January 1962 to July 2023. It reveals key trends and patterns through an innovative network analysis approach. Static network analysis is deployed to discern structural patterns in DDI research, while evolving network analysis is employed to monitor changes in the DDI research trend structures over time. Results: Our compelling results shed light on the scale-free characteristics of pharmacokinetic, pharmacodynamic, and their combined networks, exhibiting power law exponent values of 2.5, 2.82, and 2.46, respectively. In these networks, a select few drugs serve as central hubs, engaging in extensive interactions with a multitude of other drugs. Interestingly, the networks conform to a densification power law, illustrating that the number of DDIs grows exponentially as new drugs are added to the DDI network. Notably, we discovered that drugs connected in PK and PD networks predominantly belong to the same categories defined by the Anatomical Therapeutic Chemical (ATC) classification system, with fewer interactions observed between drugs from different categories. Discussion: The finding suggests that PK and PD DDIs between drugs from different ATC categories have not been studied as extensively as those between drugs within the same categories. By unearthing these hidden patterns, our study paves the way for a deeper understanding of the DDI landscape, providing valuable information for future DDI research, clinical practice, and drug development focus areas.

Analysis of age-dependent gene-expression in human tissues for studying diabetes comorbidities

The study of the relationship between type 2 diabetes mellitus (T2DM) disease and other pathologies (comorbidities), together with patient age variation, poses a challenge for medical research. There is evidence that patients affected by T2DM are more likely to develop comorbidities as they grow older. Variation of gene expression can be correlated to changes in T2DM comorbidities insurgence and progression. Understanding gene expression changes requires the analysis of large heterogeneous data at different scales as well as the integration of different data sources into network medicine models. Hence, we designed a framework to shed light on uncertainties related to age effects and comorbidity by integrating existing data sources with novel algorithms. The framework is based on integrating and analysing existing data sources under the hypothesis that changes in the basal expression of genes may be responsible for the higher prevalence of comorbidities in older patients. Using the proposed framework, we selected genes related to comorbidities from existing databases, and then analysed their expression with age at the tissues level. We found a set of genes that changes significantly in certain specific tissues over time. We also reconstructed the associated protein interaction networks and the related pathways for each tissue. Using this mechanistic framework, we detected interesting pathways related to T2DM whose genes change their expression with age. We also found many pathways related to insulin regulation and brain activities, which can be used to develop specific therapies. To the best of our knowledge, this is the first study that analyses such genes at the tissue level together with age variations.

Continuous population-level monitoring of SARS-CoV-2 seroprevalence in a large European metropolitan region

Effective public health measures against SARS-CoV-2 require granular knowledge of population-level immune responses. We developed a Tripartite Automated Blood Immunoassay (TRABI) to assess the IgG response against three SARS-CoV-2 proteins. We used TRABI for continuous seromonitoring of hospital patients and blood donors (n = 72'250) in the canton of Zurich from December 2019 to December 2020 (pre-vaccine period). We found that antibodies waned with a half-life of 75 days, whereas the cumulative incidence rose from 2.3% in June 2020 to 12.2% in mid-December 2020. A follow-up health survey indicated that about 10% of patients infected with wildtype SARS-CoV-2 sustained some symptoms at least twelve months post COVID-19. Crucially, we found no evidence of a difference in long-term complications between those whose infection was symptomatic and those with asymptomatic acute infection. The cohort of asymptomatic SARS-CoV-2-infected subjects represents a resource for the study of chronic and possibly unexpected sequelae.

The Use of a Network Analysis to Identify Associations and Temporal Patterns Among Non-communicable Diseases in Japan Based on a Large Medical Claims Database

BACKGROUND

Reducing the considerable non-communicable disease (NCD) burden in the aging Japanese population depends on better understanding of the comorbid and temporal relationships between different NCDs.

OBJECTIVE

We aimed to identify associations between NCDs and temporal patterns of NCDs in Japan using data from a large medical claims database.

METHODS

The study used three-digit International Classification of Diseases, Tenth Revision codes for NCDs for employees and their dependents included in the MinaCare database, which covers the period since 2010. Associations between pairs of NCDs were assessed by calculating risk ratios. The calculated risk ratios were used to create a network of closely associated NCDs (risk ratio > 15, statistically significant) and to assess temporal patterns of NCD diagnoses (risk ratio ≥ 5). The Infomap algorithm was used to identify clusters of diseases for different sex and age strata.

RESULTS

The analysis included 4,200,254 individuals (age < 65 years: 98%). Many of the temporal associations and patterns of the diseases of interest identified in this study were previously known. Regarding the diseases of interest, these associations can be classified as comorbidities, early manifestations initially diagnosed as something else, diseases attributable to or that cause the disease of interest, or caused by pharmacological treatment. International Classification of Diseases, Tenth Revision chapters that were most associated with other chapters included L Diseases of the skin and subcutaneous tissue. In the age-stratified and gender-stratified networks, clusters with the highest numbers of International Classification of Diseases, Tenth Revision codes included I Diseases of the circulatory system and F Mental and behavioral disorders.

CONCLUSIONS

Our findings reinforce established associations between NCDs and underline the importance of comprehensive NCD care.

Visualizing novel connections and genetic similarities across diseases using a network-medicine based approach

Understanding the genetic relationships between human disorders could lead to better treatment and prevention strategies, especially for individuals with multiple comorbidities. A common resource for studying genetic-disease relationships is the GWAS Catalog, a large and well curated repository of SNP-trait associations from various studies and populations. Some of these populations are contained within mega-biobanks such as the Million Veteran Program (MVP), which has enabled the genetic classification of several diseases in a large well-characterized and heterogeneous population. Here we aim to provide a network of the genetic relationships among diseases and to demonstrate the utility of quantifying the extent to which a given resource such as MVP has contributed to the discovery of such relations. We use a network-based approach to evaluate shared variants among thousands of traits in the GWAS Catalog repository. Our results indicate many more novel disease relationships that did not exist in early studies and demonstrate that the network can reveal clusters of diseases mechanistically related. Finally, we show novel disease connections that emerge when MVP data is included, highlighting methodology that can be used to indicate the contributions of a given biobank.

Network-medicine framework for studying disease trajectories in U.S. veterans

A better understanding of the sequential and temporal aspects in which diseases occur in patient's lives is essential for developing improved intervention strategies that reduce burden and increase the quality of health services. Here we present a network-based framework to study disease relationships using Electronic Health Records from > 9 million patients in the United States Veterans Health Administration (VHA) system. We create the Temporal Disease Network, which maps the sequential aspects of disease co-occurrence among patients and demonstrate that network properties reflect clinical aspects of the respective diseases. We use the Temporal Disease Network to identify disease groups that reflect patterns of disease co-occurrence and the flow of patients among diagnoses. Finally, we define a strategy for the identification of trajectories that lead from one disease to another. The framework presented here has the potential to offer new insights for disease treatment and prevention in large health care systems.

Network-Based Methods for Approaching Human Pathologies from a Phenotypic Point of View

Network and systemic approaches to studying human pathologies are helping us to gain insight into the molecular mechanisms of and potential therapeutic interventions for human diseases, especially for complex diseases where large numbers of genes are involved. The complex human pathological landscape is traditionally partitioned into discrete “diseases”; however, that partition is sometimes problematic, as diseases are highly heterogeneous and can differ greatly from one patient to another. Moreover, for many pathological states, the set of symptoms (phenotypes) manifested by the patient is not enough to diagnose a particular disease. On the contrary, phenotypes, by definition, are directly observable and can be closer to the molecular basis of the pathology. These clinical phenotypes are also important for personalised medicine, as they can help stratify patients and design personalised interventions. For these reasons, network and systemic approaches to pathologies are gradually incorporating phenotypic information. This review covers the current landscape of phenotype-centred network approaches to study different aspects of human diseases.

Nature and Extent of Physical Comorbidities Among Korean Patients With Mental Illnesses: Pairwise and Network Analysis Based on Health Insurance Claims Data

OBJECTIVE

The nature of physical comorbidities in patients with mental illness may differ according to diagnosis and personal characteristics. We investigated this complexity by conventional logistic regression and network analysis.

METHODS

A health insurance claims data in Korea was analyzed. For every combination of psychiatric and physical diagnoses, odds ratios were calculated adjusting age and sex. From the patient-diagnosis data, a network of diagnoses was constructed using Jaccard coefficient as the index of comorbidity.

RESULTS

In 1,017,024 individuals, 77,447 (7.6%) were diagnosed with mental illnesses. The number of physical diagnoses among them was 11.2, which was 1.6 times higher than non-psychiatric groups. The most noticeable associations were 1) neurotic illnesses with gastrointestinal/pain disorders and 2) dementia with fracture, Parkinson's disease, and cerebrovascular accidents. Unexpectedly, the diagnosis of metabolic syndrome was only scarcely found in patients with severe mental illnesses (SMIs). However, implicit associations between metabolic syndrome and SMIs were suggested in comorbidity networks.

CONCLUSION

Physical comorbidities in patients with mental illnesses were more extensive than those with other disease categories. However, the result raised questions as to whether the medical resources were being diverted to less serious conditions than more urgent conditions in patients with SMIs.

Patient-centric characterization of multimorbidity trajectories in patients with severe mental illnesses: A temporal bipartite network modeling approach

Multimorbidity is a major factor contributing to increased mortality among people with severe mental illnesses (SMI). Previous studies either focus on estimating prevalence of a disease in a population without considering relationships between diseases or ignore heterogeneity of individual patients in examining disease progression by looking merely at aggregates across a whole cohort. Here, we present a temporal bipartite network model to jointly represent detailed information on both individual patients and diseases, which allows us to systematically characterize disease trajectories from both patient and disease centric perspectives. We apply this approach to a large set of longitudinal diagnostic records for patients with SMI collected through a data linkage between electronic health records from a large UK mental health hospital and English national hospital administrative database. We find that the resulting diagnosis networks show disassortative mixing by degree, suggesting that patients affected by a small number of diseases tend to suffer from prevalent diseases. Factors that determine the network structures include an individual's age, gender and ethnicity. Our analysis on network evolution further shows that patients and diseases become more interconnected over the illness duration of SMI, which is largely driven by the process that patients with similar attributes tend to suffer from the same conditions. Our analytic approach provides a guide for future patient-centric research on multimorbidity trajectories and contributes to achieving precision medicine.

Impact of differential copayment on patient healthcare choice: evidence from South Korean National Cohort Study

OBJECTIVE

We evaluate the effectiveness of mild disease differential copayment policy aimed at reducing unnecessary patient visits to secondary/tertiary healthcare institutions in South Korea.

DESIGN

Retrospective study using difference-in-difference design.

SETTING

Sample Research database provided by the Korean National Health Insurance Service, between 2010 and 2013.

PARTICIPANTS

206 947 patients who visited healthcare institutions to treat mild diseases during the sample period.

METHODS

A linear probability model with difference-in-difference approach was adopted to estimate the changes in patients' healthcare choices associated with the differential copayment policy. The dependent variable was a binary variable denoting whether a patient visited primary healthcare or secondary/tertiary healthcare to treat her/his mild disease. Patients' individual characteristics were controlled with a fixed effect.

RESULTS

We observed significant decrease in the proportion of patients choosing secondary/tertiary healthcare over primary healthcare by 2.99 per cent point. The decrease associated with the policy was smaller by 14% in the low-income group compared with richer population, greater by 19% among the residents of Seoul metropolitan area than among people living elsewhere, and greater among frequent healthcare visitors by 33% than among people who less frequently visit healthcare.

CONCLUSION

The mild disease differential copayment policy of South Korea was successful in discouraging unnecessary visits to secondary/tertiary healthcare institutions to treat mild diseases that can be treated well in primary healthcare.

Characterizing chronological accumulation of comorbidities in healthy veterans: a computational approach

Understanding patient accumulation of comorbidities can facilitate healthcare strategy and personalized preventative care. We applied a directed network graph to electronic health record (EHR) data and characterized comorbidities in a cohort of healthy veterans undergoing screening colonoscopy. The Veterans Affairs Cooperative Studies Program #380 was a prospective longitudinal study of screening and surveillance colonoscopy. We identified initial instances of three-digit ICD-9 diagnoses for participants with at least 5 years of linked EHR history (October 1999 to December 2015). For diagnoses affecting at least 10% of patients, we calculated pairwise chronological relative risk (RR). iGraph was used to produce directed graphs of comorbidities with RR > 1, as well as summary statistics, key diseases, and communities. A directed graph based on 2210 patients visualized longitudinal development of comorbidities. Top hub (preceding) diseases included ischemic heart disease, inflammatory and toxic neuropathy, and diabetes. Top authority (subsequent) diagnoses were acute kidney failure and hypertensive chronic kidney failure. Four communities of correlated comorbidities were identified. Close analysis of top hub and authority diagnoses demonstrated known relationships, correlated sequelae, and novel hypotheses. Directed network graphs portray chronologic comorbidity relationships. We identified relationships between comorbid diagnoses in this aging veteran cohort. This may direct healthcare prioritization and personalized care.

Characterizing managing physicians by claims sequences in episodes of care

Value-based healthcare in the US is a payment structure that ties reimbursement to quality rather than volume alone. One model of value-based care is the Tennessee Division of TennCare's Episodes of Care program, which groups common health conditions into episodes using specified time windows, medical code sets and quality metrics as defined in each episode's Detailed Business Requirements [1,2]. Tennessee's program assigns responsibility for an episode to a managing physician, presenting a unique opportunity to study physician variability in cost and quality within these structured episodes. This paper proposes a pipeline for analysis demonstrated using a cohort of 599 Outpatient and Non-Acute Inpatient Cholecystectomy episodes managed by BlueCross BlueShield of Tennessee in 2016. We sorted episode claims by date of service, then calculated the pairwise Levenshtein distance between all episodes. Next, we adjusted the resulting matrix by cost dissimilarity and performed agglomerative clustering. We then examined the lowest and highest average episode cost clusters for patterns in cost and quality. Our results indicate that the facility type where the surgery takes place is important: outpatient ambulatory care center for the lowest cost cluster, and hospital operating room for the highest cost cluster. Average patient risk scores were higher in the highest cost cluster than the lowest cost cluster. Readmission rate (a quality metric tied to managing physician performance) was low for the whole cohort. Lastly, we explain how our analytical pipeline can be generalized and extended to domains beyond Episodes of Care.

Network analysis of autistic disease comorbidities in Chinese children based on ICD-10 codes

BACKGROUND

Autism is a lifelong disability associated with several comorbidities that confound diagnosis and treatment. A better understanding of these comorbidities would facilitate diagnosis and improve treatments. Our aim was to improve the detection of comorbid diseases associated with autism.

METHODS

We used an FP-growth algorithm to retrospectively infer disease associations using 1488 patients with autism treated at the Guangzhou Women and Children's Medical Center. The disease network was established using Cytoscape 3.7. The rules were internally validated by 10-fold cross-validation. All rules were further verified using the Columbia Open Health Data (COHD) and by literature search.

RESULTS

We found 148 comorbid diseases including intellectual disability, developmental speech disorder, and epilepsy. The network comprised of 76 nodes and 178 directed links. 158 links were confirmed by literature search and 105 links were validated by COHD. Furthermore, we identified 14 links not previously reported.

CONCLUSION

We demonstrate that the FP-growth algorithm can detect comorbid disease patterns, including novel ones, in patients with autism.

The architecture of co-morbidity networks of physical and mental health conditions in military veterans

Co-morbidity between medical and psychiatric conditions is commonly considered between individual pairs of conditions. However, an important alternative is to consider all conditions as part of a co-morbidity network, which encompasses all interactions between patients and a healthcare system. Analysis of co-morbidity networks could detect and quantify general tendencies not observed by smaller-scale studies. Here, we investigate the co-morbidity network derived from longitudinal healthcare records from approximately 1 million United States military veterans, a population disproportionately impacted by psychiatric morbidity and psychological trauma. Network analyses revealed marked and heterogenous patterns of co-morbidity, including a multi-scale community structure composed of groups of commonly co-morbid conditions. Psychiatric conditions including posttraumatic stress disorder were strong predictors of future medical morbidity. Neurological conditions and conditions associated with chronic pain were particularly highly co-morbid with psychiatric conditions. Across conditions, the degree of co-morbidity was positively associated with mortality. Co-morbidity was modified by biological sex and could be used to predict future diagnostic status, with out-of-sample prediction accuracy of 90-92%. Understanding complex patterns of disease co-morbidity has the potential to lead to improved designs of systems of care and the development of targeted interventions that consider the broader context of mental and physical health.

Temporal trajectories of accompanying comorbidities in patients with type 2 diabetes: a Korean nationwide observational study

Type 2 diabetes mellitus is a major concern globally and well known for increasing risk of complications. However, diabetes complications often remain undiagnosed and untreated in a large number of high-risk patients. In this study based on claims data collected in South Korea, we aimed to explore the diagnostic progression and sex- and age-related differences among patients with type 2 diabetes using time-considered patterns of the incidence of comorbidities that evolved after a diagnosis of type 2 diabetes. This study compared 164,593 patients who met the full criteria for type 2 diabetes with age group-, sex-, encounter type-, and diagnosis date-matched controls who had not been diagnosed with type 2 diabetes. We identified 76,423 significant trajectories of four diagnoses from the dataset. The top 30 trajectories with the highest average relative risks comprised microvascular, macrovascular, and miscellaneous complications. Compared with the trajectories of male groups, those of female groups included relatively fewer second-order nodes and contained hubs. Moreover, the trajectories of male groups contained diagnoses belonging to various categories. Our trajectories provide additional information about sex- and age-related differences in the risks of complications and identifying sequential relationships between type 2 diabetes and potentially complications.

High-risk multimorbidity patterns on the road to cardiovascular mortality

BACKGROUND

Multimorbidity, the co-occurrence of two or more diseases in one patient, is a frequent phenomenon. Understanding how different diseases condition each other over the lifetime of a patient could significantly contribute to personalised prevention efforts. However, most of our current knowledge on the long-term development of the health of patients (their disease trajectories) is either confined to narrow time spans or specific (sets of) diseases. Here, we aim to identify decisive events that potentially determine the future disease progression of patients.

METHODS

Health states of patients are described by algorithmically identified multimorbidity patterns (groups of included or excluded diseases) in a population-wide analysis of 9,000,000 patient histories of hospital diagnoses observed over 17 years. Over time, patients might acquire new diagnoses that change their health state; they describe a disease trajectory. We measure the age- and sex-specific risks for patients that they will acquire certain sets of diseases in the future depending on their current health state.

RESULTS

In the present analysis, the population is described by a set of 132 different multimorbidity patterns. For elderly patients, we find 3 groups of multimorbidity patterns associated with low (yearly in-hospital mortality of 0.2-0.3%), medium (0.3-1%) and high in-hospital mortality (2-11%). We identify combinations of diseases that significantly increase the risk to reach the high-mortality health states in later life. For instance, in men (women) aged 50-59 diagnosed with diabetes and hypertension, the risk for moving into the high-mortality region within 1 year is increased by the factor of 1.96 ± 0.11 (2.60 ± 0.18) compared with all patients of the same age and sex, respectively, and by the factor of 2.09 ± 0.12 (3.04 ± 0.18) if additionally diagnosed with metabolic disorders.

CONCLUSIONS

Our approach can be used both to forecast future disease burdens, as well as to identify the critical events in the careers of patients which strongly determine their disease progression, therefore constituting targets for efficient prevention measures. We show that the risk for cardiovascular diseases increases significantly more in females than in males when diagnosed with diabetes, hypertension and metabolic disorders.

Patient-Level Prediction of Cardio-Cerebrovascular Events in Hypertension Using Nationwide Claims Data

Background

Prevention and management of chronic diseases are the main goals of national health maintenance programs. Previously widely used screening tools, such as Health Risk Appraisal, are restricted in their achievement this goal due to their limitations, such as static characteristics, accessibility, and generalizability. Hypertension is one of the most important chronic diseases requiring management via the nationwide health maintenance program, and health care providers should inform patients about their risks of a complication caused by hypertension.

Objective

Our goal was to develop and compare machine learning models predicting high-risk vascular diseases for hypertensive patients so that they can manage their blood pressure based on their risk level.

Methods

We used a 12-year longitudinal dataset of the nationwide sample cohort, which contains the data of 514,866 patients and allows tracking of patients’ medical history across all health care providers in Korea (N=51,920). To ensure the generalizability of our models, we conducted an external validation using another national sample cohort dataset, comprising one million different patients, published by the National Health Insurance Service. From each dataset, we obtained the data of 74,535 and 59,738 patients with essential hypertension and developed machine learning models for predicting cardiovascular and cerebrovascular events. Six machine learning models were developed and compared for evaluating performances based on validation metrics.

Results

Machine learning algorithms enabled us to detect high-risk patients based on their medical history. The long short-term memory-based algorithm outperformed in the within test (F1-score=.772, external test F1-score=.613), and the random forest-based algorithm of risk prediction showed better performance over other machine learning algorithms concerning generalization (within test F1-score=.757, external test F1-score=.705). Concerning the number of features, in the within test, the long short-term memory-based algorithms outperformed regardless of the number of features. However, in the external test, the random forest-based algorithm was the best, irrespective of the number of features it encountered.

Conclusions

We developed and compared machine learning models predicting high-risk vascular diseases in hypertensive patients so that they may manage their blood pressure based on their risk level. By relying on the prediction model, a government can predict high-risk patients at the nationwide level and establish health care policies in advance.

Statistical methods for constructing disease comorbidity networks from longitudinal inpatient data

Tools from network science can be utilized to study relations between diseases. Different studies focus on different types of inter-disease linkages. One of them is the comorbidity patterns derived from large-scale longitudinal data of hospital discharge records. Researchers seek to describe comorbidity relations as a network to characterize pathways of disease progressions and to predict future risks. The first step in such studies is the construction of the network itself, which subsequent analyses rest upon. There are different ways to build such a network. In this paper, we provide an overview of several existing statistical approaches in network science applicable to weighted directed networks. We discuss the differences between the null models that these models assume and their applications. We apply these methods to the inpatient data of approximately one million people, spanning approximately 17 years, pertaining to the Montreal Census Metropolitan Area. We discuss the differences in the structure of the networks built by different methods, and different features of the comorbidity relations that they extract. We also present several example applications of these methods.

Association networks in a matched case-control design - Co-occurrence patterns of preexisting chronic medical conditions in patients with major depression versus their matched controls

OBJECTIVE

We present a method for comparing association networks in a matched case-control design, which provides a high-level comparison of co-occurrence patterns of features after adjusting for confounding factors. We demonstrate this approach by examining the differential distribution of chronic medical conditions in patients with major depressive disorder (MDD) compared to the distribution of these conditions in their matched controls.

MATERIALS AND METHODS

Newly diagnosed MDD patients were matched to controls based on their demographic characteristics, socioeconomic status, place of residence, and healthcare service utilization in the Korean National Health Insurance Service's National Sample Cohort. Differences in the networks of chronic medical conditions in newly diagnosed MDD cases treated with antidepressants, and their matched controls, were prioritized with a permutation test accounting for the false discovery rate. Sensitivity analyses for the associations between prioritized pairs of chronic medical conditions and new MDD diagnosis were performed with regression modeling.

RESULTS

By comparing the association networks of chronic medical conditions in newly diagnosed depression patients and their matched controls, five pairs of such conditions were prioritized among 105 possible pairs after controlling the false discovery rate at 5%. In sensitivity analyses using regression modeling, four out of the five prioritized pairs were statistically significant for the interaction terms.

CONCLUSION

Association networks in a matched case-control design can provide a high-level comparison of comorbid features after adjusting for confounding factors, thereby supplementing traditional clinical study approaches. We demonstrate the differential co-occurrence pattern of chronic medical conditions in patients with MDD and prioritize the chronic conditions that have statistically significant interactions in regression models for depression.