Learning multimorbidity patterns from electronic health records using Non-negative Matrix Factorisation

Hassaïne, Abdelâali; Canoy, Dexter; Solares, José Roberto Ayala; Rao, Shishir; Li, Yikuan; Zottoli, Mariagrazia; Rahimi, Kazem; Salimi‐Khorshidi, Gholamreza

doi:10.1016/j.jbi.2020.103606

Cited by 18 publications

(21 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The key clustering approaches used in the reviewed studies included agglomerative 24 , 31 and divisive 32 hierarchical clustering, 28 latent class analysis, 29 , 33 and graph theory, 34 all of which require substantially large health data sets. Moreover, novel machine learning approaches have been used to characterize and learn multimorbidity patterns, 35 , 36 but these methods are yet to be applied to T2DM specifically. Although the study approaches and settings were varied, predictors and predominant groups of conditions across clusters were identified and are discussed in the following sections.…”

Section: Resultsmentioning

confidence: 99%

Characterizing Multimorbidity from Type 2 Diabetes

Cicek

Buckley

Pearson‐Stuttard

et al. 2021

Endocrinology and Metabolism Clinics of North America

View full text Add to dashboard Cite

Patients with type 2 diabetes mellitus (T2DM) often live with and develop multiple co-occurring conditions, namely multimorbidity, with diffuse impacts on clinical care and patient quality of life. However, literature characterizing T2DM-related multimorbidity patterns is limited. This review summarizes the findings from the emerging literature characterizing and quantifying the association of T2DM with multimorbidity clusters. The authors’ findings reveal 3 dominant cluster types appearing in patients with T2DM-related multimorbidity, such as cardiometabolic precursor conditions, vascular conditions, and mental health conditions. The authors recommend that holistic patient care centers around early detection of other comorbidities and consideration of wider risk factors.

show abstract

Section: Resultsmentioning

confidence: 99%

Characterizing Multimorbidity from Type 2 Diabetes

Cicek

Buckley

Pearson‐Stuttard

et al. 2021

Endocrinology and Metabolism Clinics of North America

View full text Add to dashboard Cite

show abstract

“…One limitation imposed by data size is that morbidity discovery is necessarily pairwise; hence the term comorbidity, as opposed to multimorbidity discovery. Tools for multimorbidity-based discovery 43 are necessarily limited in scale due to computational constraints, considering for instance, 34 disease clusters 44 . In contrast, we have calculated pairwise comorbidities among 37,997 ICD10 diagnosis codes.…”

Section: Discussionmentioning

confidence: 99%

A Poisson binomial based statistical testing framework for comprehensive comorbidity discovery across massive Electronic Health Record datasets

Lemmon

Wesołowski

Henrie

et al. 2021

Preprint

View full text Add to dashboard Cite

Discovery of comorbidities (the concomitant occurrence of distinct medical conditions in the same patient) is a prerequisite for creating forecasting tools for downstream outcomes research. Current comorbidity discovery applications are designed for small datasets and use stratification to control for confounding variables such as age, sex, or ancestry. Stratification lowers false positive rates, but reduces power, as the size of the study cohort is decreased. Here, we describe a Poisson Binomial based approach to comorbidity discovery (PBC) designed for big-data applications that circumvents the need for stratification. PBC adjusts for confounding demographic variables on a per-patient basis, and models temporal relationships. We benchmark PBC using two datasets, the publicly available MIMIC-IV; and the entire Electronic Health Record (EHR) corpus of the University of Utah Hospital System, encompassing over 1.6 million patients, to compute comorbidity statistics on 4,623,841 pairs of potentially comorbid medical terms. The results of this computation are provided as a searchable web resource. Compared to current methods, the PBC approach reduces false positive associations, while retaining statistical power to discover true comorbidities.

show abstract

“…In the study of Hassaine et al, it was shown, using the matrix factorization method, how disease clusters progress over time, forming multimorbidity networks. An interpretation of the results is questionable, as the real clinical utility is not easy to see [107].…”

Section: The Ways To Improve Implementation Of Machine Learning/big Dmentioning

confidence: 98%

“…The presentation of temporal dynamics of multiple interactions inherent to multimorbidity is a special challenge for data scientists as it requires more sophisticated algorithms and a multi-step analytics, that go beyond a case-control classification and the use of linear regression analysis to show its progression over time [105][106][107]. The study designs for learning about temporal dynamics of this complexity are still poorly developed and are maintained within the framework of unsupervised learning (an outcome is not known) and disease-disease relationships [31,107]. Although data scientists have an absolute authority in building these sophisticated data analytics, without incorporation of domain knowledge in creation of the research task and evaluation of the interpretability of the performed analytics, the real-life usability of these innovations will be questionable [104].…”

Section: New Approaches In Multimorbidity Research Associated With Pamentioning

confidence: 99%

AI and Big Data in Healthcare: Towards a More Comprehensive Research Framework for Multimorbidity

Majnarić

Babič

O’Sullivan

et al. 2021

JCM

View full text Add to dashboard Cite

Multimorbidity refers to the coexistence of two or more chronic diseases in one person. Therefore, patients with multimorbidity have multiple and special care needs. However, in practice it is difficult to meet these needs because the organizational processes of current healthcare systems tend to be tailored to a single disease. To improve clinical decision making and patient care in multimorbidity, a radical change in the problem-solving approach to medical research and treatment is needed. In addition to the traditional reductionist approach, we propose interactive research supported by artificial intelligence (AI) and advanced big data analytics. Such research approach, when applied to data routinely collected in healthcare settings, provides an integrated platform for research tasks related to multimorbidity. This may include, for example, prediction, correlation, and classification problems based on multiple interaction factors. However, to realize the idea of this paradigm shift in multimorbidity research, the optimization, standardization, and most importantly, the integration of electronic health data into a common national and international research infrastructure is needed. Ultimately, there is a need for the integration and implementation of efficient AI approaches, particularly deep learning, into clinical routine directly within the workflows of the medical professionals.

show abstract

Learning multimorbidity patterns from electronic health records using Non-negative Matrix Factorisation

Cited by 18 publications

References 43 publications

Characterizing Multimorbidity from Type 2 Diabetes

Characterizing Multimorbidity from Type 2 Diabetes

A Poisson binomial based statistical testing framework for comprehensive comorbidity discovery across massive Electronic Health Record datasets

AI and Big Data in Healthcare: Towards a More Comprehensive Research Framework for Multimorbidity

Contact Info

Product

Resources

About