Simulation-derived best practices for clustering clinical data

Medications and cognitive decline in Alzheimer's disease: Cohort cluster analysis of 15,428 patients

BACKGROUND

Medications for comorbid conditions may affect cognition in Alzheimer's disease (AD).

OBJECTIVE

To explore the association between common medications and cognition, measured with the Mini-Mental State Examination.

METHODS

Cohort study including persons with AD from the Swedish Registry for Cognitive/Dementia Disorders (SveDem). Medications were included if they were used by ≥5% of patients (26 individual drugs). Each follow-up was analyzed independently by performing 100 Monte-Carlo simulations of two steps each 1) k-means clustering of patients according to Mini-Mental State Examination at follow-up and its decline since previous measure, and 2) Identification of medications presenting statistically significant differences in the proportion of users in the different clusters.

RESULTS

15,428 patients (60.38% women) were studied. Four clusters were identified. Medications associated with the best cognition cluster (relative to the worse) were atorvastatin (point estimate 1.44 95% confidence interval [1.15-1.83] at first follow-up, simvastatin (1.41 [1.11-1.78] at second follow-up), warfarin (1.56 [1.22-2.01] first follow-up), zopiclone (1.35 [1.15-1.58], and metformin (2.08 [1.35-3.33] second follow-up. Oxazepam (0.60 [0.50-0.73] first follow-up), paracetamol (0.83 [0.73-0.95] first follow-up), cyanocobalamin, felodipine and furosemide were associated with the worst cluster. Cholinesterase inhibitors were associated with the best cognition clusters, whereas memantine appeared in the worse cognition clusters, consistent with its indication in moderate to severe dementia.

CONCLUSIONS

We performed unsupervised clustering to classify patients based on their current cognition and cognitive decline from previous testing. Atorvastatin, simvastatin, warfarin, metformin, and zopiclone presented a positive and statistically significant associations with cognition, while oxazepam, cyanocobalamin, felodipine, furosemide and paracetamol, were associated with the worst cluster.

Artificial intelligence in pediatric allergy research

Atopic dermatitis, food allergy, allergic rhinitis, and asthma are among the most common diseases in childhood. They are heterogeneous diseases, can co-exist in their development, and manifest complex associations with other disorders and environmental and hereditary factors. Elucidating these intricacies by identifying clinically distinguishable groups and actionable risk factors will allow for better understanding of the diseases, which will enhance clinical management and benefit society and affected individuals and families. Artificial intelligence (AI) is a promising tool in this context, enabling discovery of meaningful patterns in complex data. Numerous studies within pediatric allergy have and continue to use AI, primarily to characterize disease endotypes/phenotypes and to develop models to predict future disease outcomes. However, most implementations have used relatively simplistic data from one source, such as questionnaires. In addition, methodological approaches and reporting are lacking. This review provides a practical hands-on guide for conducting AI-based studies in pediatric allergy, including (1) an introduction to essential AI concepts and techniques, (2) a blueprint for structuring analysis pipelines (from selection of variables to interpretation of results), and (3) an overview of common pitfalls and remedies. Furthermore, the state-of-the art in the implementation of AI in pediatric allergy research, as well as implications and future perspectives are discussed.

CONCLUSION

AI-based solutions will undoubtedly transform pediatric allergy research, as showcased by promising findings and innovative technical solutions, but to fully harness the potential, methodologically robust implementation of more advanced techniques on richer data will be needed.

WHAT IS KNOWN

• Pediatric allergies are heterogeneous and common, inflicting substantial morbidity and societal costs. • The field of artificial intelligence is undergoing rapid development, with increasing implementation in various fields of medicine and research.

WHAT IS NEW

• Promising applications of AI in pediatric allergy have been reported, but implementation largely lags behind other fields, particularly in regard to use of advanced algorithms and non-tabular data. Furthermore, lacking reporting on computational approaches hampers evidence synthesis and critical appraisal. • Multi-center collaborations with multi-omics and rich unstructured data as well as utilization of deep learning algorithms are lacking and will likely provide the most impactful discoveries.

Machine learning for the advancement of genome-scale metabolic modeling

Constraint-based modeling (CBM) has evolved as the core systems biology tool to map the interrelations between genotype, phenotype, and external environment. The recent advancement of high-throughput experimental approaches and multi-omics strategies has generated a plethora of new and precise information from wide-ranging biological domains. On the other hand, the continuously growing field of machine learning (ML) and its specialized branch of deep learning (DL) provide essential computational architectures for decoding complex and heterogeneous biological data. In recent years, both multi-omics and ML have assisted in the escalation of CBM. Condition-specific omics data, such as transcriptomics and proteomics, helped contextualize the model prediction while analyzing a particular phenotypic signature. At the same time, the advanced ML tools have eased the model reconstruction and analysis to increase the accuracy and prediction power. However, the development of these multi-disciplinary methodological frameworks mainly occurs independently, which limits the concatenation of biological knowledge from different domains. Hence, we have reviewed the potential of integrating multi-disciplinary tools and strategies from various fields, such as synthetic biology, CBM, omics, and ML, to explore the biochemical phenomenon beyond the conventional biological dogma. How the integrative knowledge of these intersected domains has improved bioengineering and biomedical applications has also been highlighted. We categorically explained the conventional genome-scale metabolic model (GEM) reconstruction tools and their improvement strategies through ML paradigms. Further, the crucial role of ML and DL in omics data restructuring for GEM development has also been briefly discussed. Finally, the case-study-based assessment of the state-of-the-art method for improving biomedical and metabolic engineering strategies has been elaborated. Therefore, this review demonstrates how integrating experimental and in silico strategies can help map the ever-expanding knowledge of biological systems driven by condition-specific cellular information. This multiview approach will elevate the application of ML-based CBM in the biomedical and bioengineering fields for the betterment of society and the environment.

Topological Structures in the Space of Treatment-Naïve Patients with Chronic Lymphocytic Leukemia

Patients are complex and heterogeneous; clinical data sets are complicated by noise, missing data, and the presence of mixed-type data. Using such data sets requires understanding the high-dimensional "space of patients", composed of all measurements that define all relevant phenotypes. The current state-of-the-art merely defines spatial groupings of patients using cluster analyses. Our goal is to apply topological data analysis (TDA), a new unsupervised technique, to obtain a more complete understanding of patient space. We applied TDA to a space of 266 previously untreated patients with Chronic Lymphocytic Leukemia (CLL), using the "daisy" metric to compute distances between clinical records. We found clear evidence for both loops and voids in the CLL data. To interpret these structures, we developed novel computational and graphical methods. The most persistent loop and the most persistent void can be explained using three dichotomized, prognostically important factors in CLL: IGHV somatic mutation status, beta-2 microglobulin, and Rai stage. In conclusion, patient space turns out to be richer and more complex than current models suggest. TDA could become a powerful tool in a researcher's arsenal for interpreting high-dimensional data by providing novel insights into biological processes and improving our understanding of clinical and biological data sets.

A robust clustering strategy for stratification unveils unique patient subgroups in acutely decompensated cirrhosis

BACKGROUND

Patient heterogeneity poses significant challenges for managing individuals and designing clinical trials, especially in complex diseases. Existing classifications rely on outcome-predicting scores, potentially overlooking crucial elements contributing to heterogeneity without necessarily impacting prognosis.

METHODS

To address patient heterogeneity, we developed ClustALL, a computational pipeline that simultaneously faces diverse clinical data challenges like mixed types, missing values, and collinearity. ClustALL enables the unsupervised identification of patient stratifications while filtering for stratifications that are robust against minor variations in the population (population-based) and against limited adjustments in the algorithm's parameters (parameter-based).

RESULTS

Applied to a European cohort of patients with acutely decompensated cirrhosis (n = 766), ClustALL identified five robust stratifications, using only data at hospital admission. All stratifications included markers of impaired liver function and number of organ dysfunction or failure, and most included precipitating events. When focusing on one of these stratifications, patients were categorized into three clusters characterized by typical clinical features; notably, the 3-cluster stratification showed a prognostic value. Re-assessment of patient stratification during follow-up delineated patients' outcomes, with further improvement of the prognostic value of the stratification. We validated these findings in an independent prospective multicentre cohort of patients from Latin America (n = 580).

CONCLUSIONS

By applying ClustALL to patients with acutely decompensated cirrhosis, we identified three patient clusters. Following these clusters over time offers insights that could guide future clinical trial design. ClustALL is a novel and robust stratification method capable of addressing the multiple challenges of patient stratification in most complex diseases.

SillyPutty: Improved clustering by optimizing the silhouette width

Clustering is an important task in biomedical science, and it is widely believed that different data sets are best clustered using different algorithms. When choosing between clustering algorithms on the same data set, reseachers typically rely on global measures of quality, such as the mean silhouette width, and overlook the fine details of clustering. However, the silhouette width actually computes scores that describe how well each individual element is clustered. Inspired by this observation, we developed a novel clustering method, called SillyPutty. Unlike existing methods, SillyPutty uses the silhouette width for individual elements as a tool to optimize the mean silhouette width. This shift in perspective allows for a more granular evaluation of clustering quality, potentially addressing limitations in current methodologies. To test the SillyPutty algorithm, we first simulated a series of data sets using the Umpire R package and then used real-workd data from The Cancer Genome Atlas. Using these data sets, we compared SillyPutty to several existing algorithms using multiple metrics (Silhouette Width, Adjusted Rand Index, Entropy, Normalized Within-group Sum of Square errors, and Perfect Classification Count). Our findings revealed that SillyPutty is a valid standalone clustering method, comparable in accuracy to the best existing methods. We also found that the combination of hierarchical clustering followed by SillyPutty has the best overall performance in terms of both accuracy and speed. Availability: The SillyPutty R package can be downloaded from the Comprehensive R Archive Network (CRAN).

SillyPutty: Improved clustering by optimizing the silhouette width

Unsupervised clustering is an important task in biomedical science. We developed a new clustering method, called SillyPutty, for unsupervised clustering. As test data, we generated a series of datasets using the Umpire R package. Using these datasets, we compared SillyPutty to several existing algorithms using multiple metrics (Silhouette Width, Adjusted Rand Index, Entropy, Normalized Within-group Sum of Square errors, and Perfect Classification Count). Our findings revealed that SillyPutty is a valid standalone clustering method, comparable in accuracy to the best existing methods. We also found that the combination of hierarchical clustering followed by SillyPutty has the best overall performance in terms of both accuracy and speed.

Unsupervised clustering analysis of comprehensive health status and its influencing factors on women of childbearing age: a cross-sectional study from a province in central China

BACKGROUND

Most previous studies on women of childbearing age have focused on reproductive health and fertility intentions, and evidence regarding the comprehensive health status of women of childbearing age is limited. This study aimed to comprehensively examine the health status of women of childbearing age through a multi-method and multi-indicator evaluation, analyze the factors that influence their overall health, and provide sound recommendations for the improvement and promotion of healthy behaviors.

METHODS

Data on women of childbearing age living in Shanxi Province were collected between September 2021 and January 2022 through online and offline surveys. The k-means algorithm was used to assess health-related patterns in women, and multivariate nonconditional logistic regression was used to assess the influencing factors of women's overall health.

RESULTS

In total, 1,258 of 2,925 (43%) participants were classified as having a good health status in all five domains of the three health dimensions: quality of life, mental health, and illness. Multivariate logistic regression showed that education level, gynecological examination status, health status of family members, access to medical treatment, age, cooking preferences, diet, social support, hand washing habits, attitude toward breast cancer prevention, and awareness of reproductive health were significantly associated with different health patterns.

CONCLUSIONS

The comprehensive health status of women of childbearing age in Shanxi Province is generally good; however, a large proportion of women with deficiencies in some dimensions remains. Since lifestyle greatly impacts women's health, health education on lifestyle and health-related issues should be strengthened.

Which congestion presentation pattern on the physical findings is associated with future adverse events? A cluster analysis in the multicenter acute heart failure registry

BACKGROUND

Clinical congestion is the most frequent reason for hospital admission in patients with acute heart failure (AHF). However, few studies have investigated the patterns and prognostic implication of the physical congestion using unbiased and robust statistical methods.

METHODS

A hierarchical agglomerative clustering analysis was performed in the multicenter Japanese AHF registry (N = 3151) with the distance calculated by Jaccard's distance for jugular vein distention (JVD), leg edema, S3, crackles, and orthopnea. The primary outcome was a composite of cardiac death and heart failure readmission within 1-year.

RESULTS

At the time of admission, the median number of prevalent congestive signs was 2. We identified three phenogroups: 'no physical congestions' (N = 251); 'congestion without JVD' (N = 1415); and 'congestion with JVD' (N = 1495). Patients in 'no physical congestion' were the youngest (median 75 [62, 83] years) with the lowest systolic blood pressure (122 [106, 142] mmHg). Patients in 'congestion without JVD', and 'congestion with JVD' were similar in terms of age (77 [67, 84] vs. 78 [69, 84] years) and systolic blood pressure (138 [118, 160] vs. 137 [118, 158] mmHg). While 30-day mortality was similar (4.0%, 3.7%, and 4.3% in 'no physical congestion,' 'congestion without JVD,' and 'congestion with JVD', respectively), the patients in 'congestion with JVD' were at the highest risk for the primary outcome (adjusted hazard ratio 1.79, 95% CI 1.26-2.55 when 'no physical congestion' was a reference).

CONCLUSIONS

Our clustering analysis demonstrated that congestion signs, particularly JVD, allowed identification of AHF phenogroups with distinct clinical characteristics and long-term outcomes.

K-medoids clustering of hospital admission characteristics to classify severity of influenza virus infection

Background

Patients are admitted to the hospital for respiratory illness at different stages of their disease course. It is important to appropriately analyse this heterogeneity in surveillance data to accurately measure disease severity among those hospitalized. The purpose of this study was to determine if unique baseline clusters of influenza patients exist and to examine the association between cluster membership and in-hospital outcomes.

Methods

Patients hospitalized with influenza at two hospitals in Southeast Michigan during the 2017/2018 (n = 242) and 2018/2019 (n = 115) influenza seasons were included. Physiologic and laboratory variables were collected for the first 24 h of the hospital stay. K-medoids clustering was used to determine groups of individuals based on these values. Multivariable linear regression or Firth's logistic regression were used to examine the association between cluster membership and clinical outcomes.

Results

Three clusters were selected for 2017/2018, mainly differentiated by blood glucose level. After adjustment, those in C171 had 5.6 times the odds of mechanical ventilator use than those in C172 (95% CI: 1.49, 21.1) and a significantly longer mean hospital length of stay than those in both C172 (mean 1.5 days longer, 95% CI: 0.2, 2.7) and C173 (mean 1.4 days longer, 95% CI: 0.3, 2.5). Similar results were seen between the two clusters selected for 2018/2019.

Conclusion

In this study of hospitalized influenza patients, we show that distinct clusters with higher disease acuity can be identified and could be targeted for evaluations of vaccine and influenza antiviral effectiveness against disease attenuation. The association of higher disease acuity with glucose level merits evaluation.

Risk of Early Infection in Idiopathic Inflammatory Myopathies: Cluster Analysis Based on Clinical Features and Biomarkers

Patients with idiopathic inflammatory myopathies (IIMs), referred to as myositis, are prone to infectious complications, which hinder the treatment of the disease and worsen the outcome of patients. The purpose of this study was to explore the different types of infectious complications in patients with myositis and to determine the predisposing factors for clinical reference. A retrospective study was conducted on 66 patients with IIM who were divided into different subpopulations by an unsupervised analysis of their clinical manifestations, laboratory features, and autoantibody characteristics. Combined with the incidence of infectious complications, the types of infectious pathogens and the sites of infection, the characteristics of infection, and susceptibility factors were explored. Three clusters with significantly different clinical characteristics and coinfection rates were identified (76.2% vs. 41.6% vs. 36.4%, p = 0.0139). Cluster 1 (n = 12) had a moderate risk of infection, with an infection rate of 41.6%. The patients in cluster 1 had a high probability of positive mechanic's hands, periungual erythema, anti-Ro52 antibody, and anti-Jo1 antibody. CD3 and CD4 were the highest among the three groups. Cluster 2 (n = 21) had a high risk of infection, and the incidence of infection was 76.2%. Almost all patients in this cluster had a rash, prominent clinical symptoms, and decreased WBC, PMN, LYM, CD3, and CD4 counts. Cluster 3 (n = 33) had a low risk of infection, with an infection rate of 36.4%. Compared with the other two clusters, cluster 3 (n = 33) lacked a typical rash but had a high ANA-positive rate. The patients in cluster 1 and cluster 3 were mainly infected by viruses, followed by bacterial infections. In cluster 2 patients, bacterial infections were the most prevalent. Fungal and Pneumocystis carinii were common causes of cluster 2 and 3 infections. In addition, the patients within a cluster often have a single infection, and pulmonary infections are the most common. We clustered the patients with IIM complicated with infection into three different types by their clinical symptoms and found that there were differences in the infection risk and infection types among the different cluster groups.

Using Unsupervised Machine Learning Methods to Cluster Comorbidities in a Population-Based Cohort of Patients With Rheumatoid Arthritis

OBJECTIVE

To identify clusters of comorbidities in patients with rheumatoid arthritis (RA) using 4 methods and to compare to patients without RA.

METHODS

In this retrospective, population-based study, residents of 8 Minnesota counties with prevalent RA as of January 1, 2015 were identified. Age-, sex-, and county-matched non-RA comparators were selected from the same underlying population. Diagnostic codes were retrieved for 5 years before January 1, 2015. Using 2 codes ≥30 days apart, 44 previously defined morbidities and 11 nonoverlapping chronic disease categories based on Clinical Classifications Software were defined. Unsupervised machine learning methods of interest included hierarchical clustering, factor analysis, K-means clustering, and network analysis.

RESULTS

Two groups of 1,643 patients with and without RA (72% female; mean age 63.1 years in both groups) were studied. Clustering of comorbidities revealed strong associations among mental/behavioral comorbidities and among cardiovascular risk factors and diseases. The clusters were associated with age and sex. Differences between the 4 clustering methods were driven by comorbidities that are rare and those that were weakly associated with other comorbidities. Common comorbidities tended to group together consistently across approaches. The instability of clusters when using different random seeds or bootstrap sampling impugns the usefulness and reliability of these methods. Clusters of common comorbidities between RA and non-RA cohorts were similar.

CONCLUSION

Despite the higher comorbidity burden in patients with RA compared to the general population, clustering comorbidities did not identify substantial differences in comorbidity patterns between the RA and non-RA cohorts. The instability of clustering methods suggests caution when interpreting clustering using 1 method.

A cohesin-associated gene score may predict immune checkpoint blockade in hepatocellular carcinoma

Stromal antigen 1 (STAG1), a component of cohesion, is overexpressed in various cancers, but it is unclear whether it has a role in the transcriptional regulation of hepatocellular carcinoma (HCC). To test this hypothesis, here, we screened global HCC datasets and performed multiscale embedded gene co-expression network analysis to identify the potential functional modules of differentially expressed STAG1 co-expressed genes. The putative transcriptional targets of STAG1 were identified using chromatin immunoprecipitation followed by high-throughput DNA sequencing. The cohesin-associated gene score (CAGS) was quantified using the The Cancer Genome Atlas HCC cohort and single-sample gene set enrichment analysis. Distinct cohesin-associated gene patterns were identified by calculating the euclidean distance of each patient. We assessed the potential ability of the CAGS in predicting immune checkpoint blockade (ICB) treatment response using IMvigor210 and GSE78220 cohorts. STAG1 was upregulated in 3313 HCC tissue samples compared with 2692 normal liver tissue samples (standard mean difference = 0.54). A total of three cohesin-associated gene patterns were identified, where cluster 2 had a high TP53 mutated rate and a poor survival outcome. Low CAGS predicted a significant survival advantage but presaged poor immunotherapy response. Differentially expressed STAG1 co-expression genes were enriched in the mitotic cell cycle, lymphocyte activation, and blood vessel development. PDS5A and PDGFRA were predicted as the downstream transcriptional targets of STAG1. In summary, STAG1 is significantly upregulated in global HCC tissue samples and may participate in blood vessel development and the mitotic cell cycle. A cohesin-associated gene scoring system may have potential to predict the ICB response.