Network Analysis of Human Disease Comorbidity Patterns Based on Large-Scale Data Mining

"Using network analysis modularity to group health code systems and decrease dimensionality in machine learning models"

Background

Machine learning (ML) prediction models in healthcare and pharmacy-related research face challenges with encoding high-dimensional Healthcare Coding Systems (HCSs) such as ICD, ATC, and DRG codes, given the trade-off between reducing model dimensionality and minimizing information loss.

Objectives

To investigate using Network Analysis modularity as a method to group HCSs to improve encoding in ML models.

Methods

The MIMIC-III dataset was utilized to create a multimorbidity network in which ICD-9 codes are the nodes and the edges are the number of patients sharing the same ICD-9 code pairs. A modularity detection algorithm was applied using different resolution thresholds to generate 6 sets of modules. The impact of four grouping strategies on the performance of predicting 90-day Intensive Care Unit readmissions was assessed. The grouping strategies compared: 1) binary encoding of codes, 2) encoding codes grouped by network modules, 3) grouping codes to the highest level of ICD-9 hierarchy, and 4) grouping using the single-level Clinical Classification Software (CCS). The same methodology was also applied to encode DRG codes but limiting the comparison to a single modularity threshold to binary encoding.The performance was assessed using Logistic Regression, Support Vector Machine with a non-linear kernel, and Gradient Boosting Machines algorithms. Accuracy, Precision, Recall, AUC, and F1-score with 95% confidence intervals were reported.

Results

Models utilized modularity encoding outperformed ungrouped codes binary encoding models. The accuracy improved across all algorithms ranging from 0.736 to 0.78 for the modularity encoding, to 0.727 to 0.779 for binary encoding. AUC, recall, and precision also improved across almost all algorithms. In comparison with other grouping approaches, modularity encoding generally showed slightly higher performance in AUC, ranging from 0.813 to 0.837, and precision, ranging from 0.752 to 0.782.

Conclusions

Modularity encoding enhances the performance of ML models in pharmacy research by effectively reducing dimensionality and retaining necessary information. Across the three algorithms used, models utilizing modularity encoding showed superior or comparable performance to other encoding approaches. Modularity encoding introduces other advantages such as it can be used for both hierarchical and non-hierarchical HCSs, the approach is clinically relevant, and can enhance ML models' clinical interpretation. A Python package has been developed to facilitate the use of the approach for future research.

Analysis of disease comorbidity patterns in a large-scale China population

BACKGROUND

Disease comorbidity is popular and has significant indications for disease progress and management. We aim to detect the general disease comorbidity patterns in Chinese populations using a large-scale clinical data set.

METHODS

We extracted the diseases from a large-scale anonymized data set derived from 8,572,137 inpatients in 453 hospitals across China. We built a Disease Comorbidity Network (DCN) using correlation analysis and detected the topological patterns of disease comorbidity using both complex network and data mining methods. The comorbidity patterns were further validated by shared molecular mechanisms using disease-gene associations and pathways. To predict the disease occurrence during the whole disease progressions, we applied four machine learning methods to model the disease trajectories of patients.

RESULTS

We obtained the DCN with 5702 nodes and 258,535 edges, which shows a power law distribution of the degree and weight. It further indicated that there exists high heterogeneity of comorbidities for different diseases and we found that the DCN is a hierarchical modular network with community structures, which have both homogeneous and heterogeneous disease categories. Furthermore, adhering to the previous work from US and Europe populations, we found that the disease comorbidities have their shared underlying molecular mechanisms. Furthermore, take hypertension and psychiatric disease as instance, we used four classification methods to predicte the disease occurrence using the comorbid disease trajectories and obtained acceptable performance, in which in particular, random forest obtained an overall best performance (with F1-score 0.6689 for hypertension and 0.6802 for psychiatric disease).

CONCLUSIONS

Our study indicates that disease comorbidity is significant and valuable to understand the disease incidences and their interactions in real-world populations, which will provide important insights for detection of the patterns of disease classification, diagnosis and prognosis.

Applications of network analysis to routinely collected health care data: a systematic review

Objective

To survey network analyses of datasets collected in the course of routine operations in health care settings and identify driving questions, methods, needs, and potential for future research.

Materials and Methods

A search strategy was designed to find studies that applied network analysis to routinely collected health care datasets and was adapted to 3 bibliographic databases. The results were grouped according to a thematic analysis of their settings, objectives, data, and methods. Each group received a methodological synthesis.

Results

The search found 189 distinct studies reported before August 2016. We manually partitioned the sample into 4 groups, which investigated institutional exchange, physician collaboration, clinical co-occurrence, and workplace interaction networks. Several robust and ongoing research programs were discerned within (and sometimes across) the groups. Little interaction was observed between these programs, despite conceptual and methodological similarities.

Discussion

We use the literature sample to inform a discussion of good practice at this methodological interface, including the concordance of motivations, study design, data, and tools and the validation and standardization of techniques. We then highlight instances of positive feedback between methodological development and knowledge domains and assess the overall cohesion of the sample.

Computer technologies to integrate medical treatments to manage multimorbidity

The high prevalence of multimorbid cases is a challenge for Health-Care Systems today. Clinical practice guidelines are the means to register and transmit the available evidence-based medical knowledge concerning concrete diseases. Several computer languages have been defined to represent this knowledge in a way that computers could use to help physicians in the daily practice of medicine. The generation of guidelines for all possible multimorbidities entails several issues that are difficult to address. Consequently, numerous medical informatics technologies have appeared merging computer information structures in a way that the treatment knowledge about single diseases could be combined in order to deliver health-care to patients suffering from multimorbidity. This paper proposes a classification of the most promising current technologies addressing this issue and provides an analysis of their maturity, strengths, and weaknesses. We conclude with an enumeration of ten relevant issues to consider when developing such technologies.

Phenome-driven disease genetics prediction toward drug discovery

MOTIVATION

Discerning genetic contributions to diseases not only enhances our understanding of disease mechanisms, but also leads to translational opportunities for drug discovery. Recent computational approaches incorporate disease phenotypic similarities to improve the prediction power of disease gene discovery. However, most current studies used only one data source of human disease phenotype. We present an innovative and generic strategy for combining multiple different data sources of human disease phenotype and predicting disease-associated genes from integrated phenotypic and genomic data.

RESULTS

To demonstrate our approach, we explored a new phenotype database from biomedical ontologies and constructed Disease Manifestation Network (DMN). We combined DMN with mimMiner, which was a widely used phenotype database in disease gene prediction studies. Our approach achieved significantly improved performance over a baseline method, which used only one phenotype data source. In the leave-one-out cross-validation and de novo gene prediction analysis, our approach achieved the area under the curves of 90.7% and 90.3%, which are significantly higher than 84.2% (P < e(-4)) and 81.3% (P < e(-12)) for the baseline approach. We further demonstrated that our predicted genes have the translational potential in drug discovery. We used Crohn's disease as an example and ranked the candidate drugs based on the rank of drug targets. Our gene prediction approach prioritized druggable genes that are likely to be associated with Crohn's disease pathogenesis, and our rank of candidate drugs successfully prioritized the Food and Drug Administration-approved drugs for Crohn's disease. We also found literature evidence to support a number of drugs among the top 200 candidates. In summary, we demonstrated that a novel strategy combining unique disease phenotype data with system approaches can lead to rapid drug discovery.

AVAILABILITY AND IMPLEMENTATION

nlp.

CASE

edu/public/data/DMN