Artificial intelligence driven definition of food preference endotypes in UK Biobank volunteers is associated with distinctive health outcomes and blood based metabolomic and proteomic profiles.

BACKGROUND: Specific food preferences can determine an individual's dietary patterns and therefore, may be associated with certain health risks and benefits. METHODS: Using food preference questionnaire (FPQ) data from a subset comprising over 180,000 UK Biobank participants, we employed Latent Profile Analysis (LPA) approach to identify the main patterns or profiles among participants. blood biochemistry across groups/profiles was compared using the non-parametric Kruskal-Wallis test. We applied the Limma algorithm for differential abundance analysis on 168 metabolites and 2923 proteins, and utilized the Database for Annotation, Visualization and Integrated Discovery (DAVID) to identify enriched biological processes and pathways. Relative risks (RR) were calculated for chronic diseases and mental conditions per group, adjusting for sociodemographic factors. RESULTS: Based on their food preferences, three profiles were termed: the putative Health-conscious group (low preference for animal-based or sweet foods, and high preference for vegetables and fruits), the Omnivore group (high preference for all foods), and the putative Sweet-tooth group (high preference for sweet foods and sweetened beverages). The Health-conscious group exhibited lower risk of heart failure (RR = 0.86, 95%CI 0.79-0.93) and chronic kidney disease (RR = 0.69, 95%CI 0.65-0.74) compared to the two other groups. The Sweet-tooth group had greater risk of depression (RR = 1.27, 95%CI 1.21-1.34), diabetes (RR = 1.15, 95%CI 1.01-1.31), and stroke (RR = 1.22, 95%CI 1.15-1.31) compared to the other two groups. Cancer (overall) relative risk showed little difference across the Health-conscious, Omnivore, and Sweet-tooth groups with RR of 0.98 (95%CI 0.96-1.01), 1.00 (95%CI 0.98-1.03), and 1.01 (95%CI 0.98-1.04), respectively. The Health-conscious group was associated with lower levels of inflammatory biomarkers (e.g., C-reactive Protein) which are also known to be elevated in those with common metabolic diseases (e.g., cardiovascular disease). Other markers modulated in the Health-conscious group, ketone bodies, insulin-like growth factor-binding protein (IGFBP), and Growth Hormone 1 were more abundant, while leptin was less abundant. Further, the IGFBP pathway, which influences IGF1 activity, may be significantly enhanced by dietary choices. CONCLUSIONS: These observations align with previous findings from studies focusing on weight loss interventions, which include a reduction in leptin levels. Overall, the Health-conscious group, with preference to healthier food options, has better health outcomes, compared to Sweet-tooth and Omnivore groups.

Derivation and validation of generalized sepsis-induced acute respiratory failure phenotypes among critically ill patients: a retrospective study.

BACKGROUND: Septic patients who develop acute respiratory failure (ARF) requiring mechanical ventilation represent a heterogenous subgroup of critically ill patients with widely variable clinical characteristics. Identifying distinct phenotypes of these patients may reveal insights about the broader heterogeneity in the clinical course of sepsis, considering multi-organ dynamics. We aimed to derive novel phenotypes of sepsis-induced ARF using observational clinical data and investigate the generalizability of the derived phenotypes. METHODS: We performed a multi-center retrospective study of ICU patients with sepsis who required mechanical ventilation for ≥ 24 h. Data from two different high-volume academic hospital centers were used, where all phenotypes were derived in MICU of Hospital-I (N = 3225). The derived phenotypes were validated in MICU of Hospital-II (N = 848), SICU of Hospital-I (N = 1112), and SICU of Hospital-II (N = 465). Clinical data from 24 h preceding intubation was used to derive distinct phenotypes using an explainable machine learning-based clustering model interpreted by clinical experts. RESULTS: Four distinct ARF phenotypes were identified: A (severe multi-organ dysfunction (MOD) with a high likelihood of kidney injury and heart failure), B (severe hypoxemic respiratory failure [median P/F = 123]), C (mild hypoxia [median P/F = 240]), and D (severe MOD with a high likelihood of hepatic injury, coagulopathy, and lactic acidosis). Patients in each phenotype showed differences in clinical course and mortality rates despite similarities in demographics and admission co-morbidities. The phenotypes were reproduced in external validation utilizing the MICU of Hospital-II and SICUs from Hospital-I and -II. Kaplan-Meier analysis showed significant difference in 28-day mortality across the phenotypes (p < 0.01) and consistent across MICU and SICU of both Hospital-I and -II. The phenotypes demonstrated differences in treatment effects associated with high positive end-expiratory pressure (PEEP) strategy. CONCLUSION: The phenotypes demonstrated unique patterns of organ injury and differences in clinical outcomes, which may help inform future research and clinical trial design for tailored management strategies.

Utilizing echocardiography and unsupervised machine learning for heart failure risk identification.

BACKGROUND: Global longitudinal strain (GLS) is recognized as a powerful predictor of heart failure (HF). However, the entire strain curve may entail important prognostic information regarding HF risk that might be undiscovered by only focusing on the peak strain value. OBJECTIVE: The hypothesis of the present study was, that analysis of the entire strain curve using unsupervised machine learning (uML) would reveal novel ventricular deformation patterns capable of predicting incident HF independently of GLS. METHODS: Longitudinal strain curves from 3710 subjects from the general population without prevalent HF were analyzed using uML. RESULTS: Mean age was 56 years and 43 % were male. During a median follow-up of 5.3 years, 92 subjects (2.5 %) developed HF. The uML algorithm generated a hierarchical clustering tree (HCT) resulting in 10 different clusters. Generally, the strain curves displayed reduced early diastolic strain to peak-strain ratio with an increasing incidence rate of HF. In multivariable Cox regressions, cluster 9 was significantly associated with increased risk of HF when compared to cluster 2-5, and 7-8 [For cluster 3: HR 8.95, 95 %CI: 2.08;38.48, P = 0.003] even though the subjects of cluster 9 were younger, displayed healthier clinical baseline characteristics, and only had slightly reduced GLS. The mean strain curve of cluster 9 displayed an early systolic lengthening followed by a late and reduced contraction specifically related to the basal lateral segment. CONCLUSION: The unsupervised machine learning algorithm identified unknown strain patterns beyond GLS presumably related to increased risk of HF.

Unsupervised machine learning for identifying attention-deficit/hyperactivity disorder subtypes based on cognitive function and their implications for brain structure.

BACKGROUND: Structural anomalies in the frontal lobe and basal ganglia have been reported in patients with attention-deficit/hyperactivity disorder (ADHD). However, these findings have been not always consistent because of ADHD diversity. This study aimed to identify ADHD subtypes based on cognitive function and find their distinct brain structural characteristics. METHODS: Using the data of 656 children with ADHD from the Adolescent Brain Cognitive Development (ABCD) Study, we applied unsupervised machine learning to identify ADHD subtypes using the National Institutes of Health Toolbox Tasks. Moreover, we compared the regional brain volumes between each ADHD subtype and 6601 children without ADHD (non-ADHD). RESULTS: Hierarchical cluster analysis automatically classified ADHD into three distinct subtypes: ADHD-A (n = 212, characterized by high-order cognitive ability), ADHD-B (n = 190, characterized by low cognitive control, processing speed, and episodic memory), and ADHD-C (n = 254, characterized by strikingly low cognitive control, working memory, episodic memory, and language ability). Structural analyses revealed that the ADHD-C type had significantly smaller volumes of the left inferior temporal gyrus and right lateral orbitofrontal cortex than the non-ADHD group, and the right lateral orbitofrontal cortex volume was positively correlated with language performance in the ADHD-C type. However, the volumes of the ADHD-A and ADHD-B types were not significantly different from those of the non-ADHD group. CONCLUSIONS: These results indicate the presence of anomalies in the lateral orbitofrontal cortex associated with language deficits in the ADHD-C type. Subtype specificity may explain previous inconsistencies in brain structural anomalies reported in ADHD.

Dissonance between posts of health agencies and public comments regarding COVID-19 and vaccination on Facebook in Northern California.

BACKGROUND: Public health crises, such as the COVID-19 pandemic, have prompted a need for health agencies to improve their disease preparedness strategies, informing their communities of new information and promoting preventive behaviors to help curb the spread of the virus. METHODS: We ran unsupervised machine learning and emotion analysis, validated with manual coding, on posts of health agencies (N = 1588) and their associated public comments (N = 7813) during a crucial initial period of the COVID-19 pandemic (January 2020 to February 2021) among nine different counties with a higher proportion of vaccine-hesitant communities in Northern California. In addition, we explored differences in concerns and expressed emotions by two key group-level factors, county-level COVID-19 death rate and political party affiliation. RESULTS: We consistently find that while health agencies primarily disseminated information about COVID-19 and the vaccine, they failed to address the concerns of their communities as expressed in public comment sections. Topics among public audiences focused on concerns with the COVID-19 vaccine safety and rollout, state mandates, flu vaccination, and frustration with politicians, and they expressed more positive and more negative emotions than health agencies. Further, there were several differences in primary topics and emotions expressed among public audiences by county-level COVID-19 death rate and political party affiliation. CONCLUSION: While this research serves as a case study, findings indicate how local health agencies, and their audiences, discuss their perceptions and concerns regarding the COVID-19 pandemic and may inform health communication researchers and practitioners on how to prepare and manage for emerging health crises.

SMS Scam Detection Application Based on Optical Character Recognition for Image Data Using Unsupervised and Deep Semi-Supervised Learning.

The growing problem of unsolicited text messages (smishing) and data irregularities necessitates stronger spam detection solutions. This paper explores the development of a sophisticated model designed to identify smishing messages by understanding the complex relationships among words, images, and context-specific factors, areas that remain underexplored in existing research. To address this, we merge a UCI spam dataset of regular text messages with real-world spam data, leveraging OCR technology for comprehensive analysis. The study employs a combination of traditional machine learning models, including K-means, Non-Negative Matrix Factorization, and Gaussian Mixture Models, along with feature extraction techniques such as TF-IDF and PCA. Additionally, deep learning models like RNN-Flatten, LSTM, and Bi-LSTM are utilized. The selection of these models is driven by their complementary strengths in capturing both the linear and non-linear relationships inherent in smishing messages. Machine learning models are chosen for their efficiency in handling structured text data, while deep learning models are selected for their superior ability to capture sequential dependencies and contextual nuances. The performance of these models is rigorously evaluated using metrics like accuracy, precision, recall, and F1 score, enabling a comparative analysis between the machine learning and deep learning approaches. Notably, the K-means feature extraction with vectorizer achieved 91.01% accuracy, and the KNN-Flatten model reached 94.13% accuracy, emerging as the top performer. The rationale behind highlighting these models is their potential to significantly improve smishing detection rates. For instance, the high accuracy of the KNN-Flatten model suggests its applicability in real-time spam detection systems, but its computational complexity might limit scalability in large-scale deployments. Similarly, while K-means with vectorizer excels in accuracy, it may struggle with the dynamic and evolving nature of smishing attacks, necessitating continual retraining.

Unsupervised discovery of family specific vocal usage in the Mongolian gerbil.

In nature, animal vocalizations can provide crucial information about identity, including kinship and hierarchy. However, lab-based vocal behavior is typically studied during brief interactions between animals with no prior social relationship, and under environmental conditions with limited ethological relevance. Here, we address this gap by establishing long-term acoustic recordings from Mongolian gerbil families, a core social group that uses an array of sonic and ultrasonic vocalizations. Three separate gerbil families were transferred to an enlarged environment and continuous 20-day audio recordings were obtained. Using a variational autoencoder (VAE) to quantify 583,237 vocalizations, we show that gerbils exhibit a more elaborate vocal repertoire than has been previously reported and that vocal repertoire usage differs significantly by family. By performing gaussian mixture model clustering on the VAE latent space, we show that families preferentially use characteristic sets of vocal clusters and that these usage preferences remain stable over weeks. Furthermore, gerbils displayed family-specific transitions between vocal clusters. Since gerbils live naturally as extended families in complex underground burrows that are adjacent to other families, these results suggest the presence of a vocal dialect which could be exploited by animals to represent kinship. These findings position the Mongolian gerbil as a compelling animal model to study the neural basis of vocal communication and demonstrates the potential for using unsupervised machine learning with uninterrupted acoustic recordings to gain insights into naturalistic animal behavior.

Machine Learning for Deconvolution and Segmentation of Hyperspectral Imaging Data from Biopharmaceutical Resins.

Biopharmaceutical resins are pivotal inert matrices used across industry and academia, playing crucial roles in a myriad of applications. For biopharmaceutical process research and development applications, a deep understanding of the physical and chemical properties of the resin itself is frequently required, including for drug purification, drug delivery, and immobilized biocatalysis. Nevertheless, the prevailing methodologies currently employed for elucidating these important aspects of biopharmaceutical resins are often lacking, frequently require significant sample alteration, are destructive or ionizing in nature, and may not adequately provide representative information. In this work, we propose the use of unsupervised machine learning technologies, in the form of both non-negative matrix factorization (NMF) and k-means segmentation, in conjugation with Raman hyperspectral imaging to rapidly elucidate the molecular and spatial properties of biopharmaceutical resins. Leveraging our proposed technology, we offer a new approach to comprehensively understanding important resin-based systems for application across biopharmaceuticals and beyond. Specifically, focusing herein on a representative resin widely utilized across the industry (i.e., Immobead 150P), our findings showcase the ability of our machine learning-based technology to molecularly identify and spatially resolve all chemical species present. Further, we offer a comprehensive evaluation of optimal excitation for hyperspectral imaging data collection, demonstrating results across 532, 638, and 785 nm excitation. In all cases, our proposed technology deconvoluted, both spatially and spectrally, resin and glass substrates via NMF. After NMF deconvolution, image segmentation was also successfully accomplished in all data sets via k-means clustering. To the best of our knowledge, this is the first report utilizing the combination of two unsupervised machine learning methodologies, combining NMF and k-means, for the rapid deconvolution and segmentation of biopharmaceutical resins. As such, we offer a powerful new data-rich experimentation tool for application across multidisciplinary fields for a deeper understanding of resins.

Enhancing pharmacist intervention targeting based on patient clustering with unsupervised machine learning.

OBJECTIVES: Adherence to the American Diabetes Association (ADA) Standards of Medical Care is low. This study aimed to assist pharmacists in identifying patients for diabetes control interventions using unsupervised machine learning. METHODS: This study analyzed the 2021 Medical Expenditure Panel Survey and used a k-mode cluster analysis. Patient features analyzed were adherence to a select set of preventive measures from the ADA Standards of Medical Care (HbA1c test, foot examination, blood cholesterol test, dilated eye examination, and influenza vaccination) and some patient characteristics (age, gender, health insurance, insulin use, and diabetes-related complications). RESULTS: The study included 1,219 patients with self-reported diabetes, and the adherence rate to the ADA standards was 33.72%. Five distinct clusters emerged: (A) moderate-complexity, privately insured male; (B) moderate-complexity, publicly insured female; (C) low-complexity, privately insured female; (D) high-complexity, publicly insured female; (E) moderate-complexity, publicly insured male. Groups B, C, and E exhibited nonadherence. CONCLUSIONS: Pharmacists can target publicly insured elderly (Groups B and E) and privately insured middle-aged females (Group C) for interventions. For instance, pharmacists may help patients in Groups B and E locate existing resources in their insurance program and remind those in Group C of the importance of adequate diabetes care.

Unsupervised Machine Learning to Identify Risk Factors of Pyeloplasty Failure in Ureteropelvic Junction Obstruction.

Introduction: In adult patients with ureteropelvic junction obstruction (UPJO), little data exist on predicting pyeloplasty outcome, and there is no unified definition of pyeloplasty success. As such, defining pyeloplasty success retrospectively is particularly vulnerable to bias, allowing researchers to choose significant outcomes with the benefit of hindsight. To mitigate these biases, we performed an unsupervised machine learning cluster analysis on a dataset of 216 pyeloplasty patients between 2015 and 2023 from a multihospital system to identify the defining risk factors of patients that experience worse outcomes. Methods: A KPrototypes model was fitted with pre- and perioperative data and blinded to postoperative outcomes. T-test and chi-square tests were performed to look at significant differences of characteristics between clusters. SHapley Additive exPlanation values were calculated from a random forest classifier to determine the most predictive features of cluster membership. A logistic regression model identified which of the most predictive variables remained significant after adjusting for confounding effects. Results: Two distinct clusters were identified. One cluster (denoted as "high-risk") contained 111 (51.4%) patients and was identified by having more comorbidities, such as old age (62.7 vs 35.7), high body mass index (BMI) (26.9 vs 23.8), hypertension (66.7% vs 17.1%), and previous abdominal surgery (72.1% vs 37.1%) and was found to have worse outcomes, such as more frequent severe postoperative complications (7.2% vs 1.0%). After adjusting for confounding effects, the most predictive features of high-risk cluster membership were old age, low preoperative estimated glomerular filtration rate (eGFR), hypertension, greater BMI, previous abdominal surgery, and left-sided UPJO. Conclusions: Adult UPJO patients with older age, lower eGFR, hypertension, greater BMI, previous abdominal surgery, and left-sided UPJO naturally cluster into to a group that more commonly suffers from perioperative complications and worse outcomes. Preoperative counseling and perioperative management for patients with these risk factors may need to be thought of or approached differently.

Machine learning-assisted rapid determination for traditional Chinese Medicine Constitution.

The aim of this study was to develop a machine learning-assisted rapid determination methodology for traditional Chinese Medicine Constitution. Based on the Constitution in Chinese Medicine Questionnaire (CCMQ), the most applied diagnostic instrument for assessing individuals' constitutions, we employed automated supervised machine learning algorithms (i.e., Tree-based Pipeline Optimization Tool; TPOT) on all the possible item combinations for each subscale and an unsupervised machine learning algorithm (i.e., variable clustering; varclus) on the whole scale to select items that can best predict body constitution (BC) classifications or BC scores. By utilizing subsets of items selected based on TPOT and corresponding machine learning algorithms, the accuracies of BC classifications prediction ranged from 0.819 to 0.936, with the root mean square errors of BC scores prediction stabilizing between 6.241 and 9.877. Overall, the results suggested that the automated machine learning algorithms performed better than the varclus algorithm for item selection. Additionally, based on an automated machine learning item selection procedure, we provided the top three ranked item combinations with each possible subscale length, along with their corresponding algorithms for predicting BC classification and severity. This approach could accommodate the needs of different practitioners in traditional Chinese medicine for rapid constitution determination.

Unsupervised machine learning highlights the challenges of subtyping disorders of gut-brain interaction.

BACKGROUND: Unsupervised machine learning describes a collection of powerful techniques that seek to identify hidden patterns in unlabeled data. These techniques can be broadly categorized into dimension reduction, which transforms and combines the original set of measurements to simplify data, and cluster analysis, which seeks to group subjects based on some measure of similarity. Unsupervised machine learning can be used to explore alternative subtyping of disorders of gut-brain interaction (DGBI) compared to the existing gastrointestinal symptom-based definitions of Rome IV. PURPOSE: This present review aims to familiarize the reader with fundamental concepts of unsupervised machine learning using accessible definitions and provide a critical summary of their application to the evaluation of DGBI subtyping. By considering the overlap between Rome IV clinical definitions and identified clusters, along with clinical and physiological insights, this paper speculates on the possible implications for DGBI. Also considered are algorithmic developments in the unsupervised machine learning community that may help leverage increasingly available omics data to explore biologically informed definitions. Unsupervised machine learning challenges the modern subtyping of DGBI and, with the necessary clinical validation, has the potential to enhance future iterations of the Rome criteria to identify more homogeneous, diagnosable, and treatable patient populations.

Phenotypes of Patients with Intracerebral Hemorrhage, Complications, and Outcomes.

BACKGROUND: The objective of this study was to define clinically meaningful phenotypes of intracerebral hemorrhage (ICH) using machine learning. METHODS: We used patient data from two US medical centers and the Antihypertensive Treatment of Acute Cerebral Hemorrhage-II clinical trial. We used k-prototypes to partition patient admission data. We then used silhouette method calculations and elbow method heuristics to optimize the clusters. Associations between phenotypes, complications (e.g., seizures), and functional outcomes were assessed using the Kruskal-Wallis H-test or χ2 test. RESULTS: There were 916 patients; the mean age was 63.8 ± 14.1 years, and 426 patients were female (46.5%). Three distinct clinical phenotypes emerged: patients with small hematomas, elevated blood pressure, and Glasgow Coma Scale scores > 12 (n = 141, 26.6%); patients with hematoma expansion and elevated international normalized ratio (n = 204, 38.4%); and patients with median hematoma volumes of 24 (interquartile range 8.2-59.5) mL, who were more frequently Black or African American, and who were likely to have intraventricular hemorrhage (n = 186, 35.0%). There were associations between clinical phenotype and seizure (P = 0.024), length of stay (P = 0.001), discharge disposition (P < 0.001), and death or disability (modified Rankin Scale scores 4-6) at 3-months' follow-up (P < 0.001). We reproduced these three clinical phenotypes of ICH in an independent cohort (n = 385) for external validation. CONCLUSIONS: Machine learning identified three phenotypes of ICH that are clinically significant, associated with patient complications, and associated with functional outcomes. Cerebellar hematomas are an additional phenotype underrepresented in our data sources.

Predictive model for novel subtypes of patients undergoing lower extremity amputation for peripheral artery disease: An unsupervised machine learning study.

Background: Peripheral artery disease (PAD) represents the frequently seen circulatory condition related to a risk of critical limb ischemia and amputation. Critical lower extremity ischemia may require amputation, and the outcomes vary. In this study, we developed an artificial intelligence (AI)-driven predictive model for PAD subtypes to assess risk among patients more precisely and accurately to predict disease progression. Methods: The present retrospective study examined clinical data in PAD patents undergoing lower extremity amputation. The data were analyzed using an unsupervised machine learning algorithm (UMLA) for subgroup identification and risk stratification. The clustering result accuracy was validated by analyzing the follow-up data of clusters. Finally, we built the prediction model with binary logistic regression. Results: In total, we enrolled 507 cases into this work. Two distinct subgroups, consisting of Clusters 1 and 2, were identified by UMLA; those from Cluster 1 showed markedly poorer conditions and prognostic outcomes compared with those from Cluster 2. With regard to the new PAD subtype, we established a nomogram with eight predictive factors, including gender, age, smoking history, diabetes and coronary heart disease history, albumin levels, endovascular intervention, and amputation level. The nomogram could accurately categorize patients into two identified clusters, and the area under receiver operating characteristic curve was 0.861 (95 % confidence interval: 0.830-0.893). Conclusion: In this study, UMLA was used to identify new phenotypic subgroups among PAD cases who showed different risks of amputation. Our constructed AI-driven predictive model for PAD subtypes showed that it can be used for risk stratification and clinical management with high accuracy and reliability.

Sequence-based detection of emerging antigenically novel influenza A viruses.

The detection of evolutionary transitions in influenza A (H3N2) viruses' antigenicity is a major obstacle to effective vaccine design and development. In this study, we describe Novel Influenza Virus A Detector (NIAViD), an unsupervised machine learning tool, adept at identifying these transitions, using the HA1 sequence and associated physico-chemical properties. NIAViD performed with 88.9% (95% CI, 56.5-98.0%) and 72.7% (95% CI, 43.4-90.3%) sensitivity in training and validation, respectively, outperforming the uncalibrated null model-33.3% (95% CI, 12.1-64.6%) and does not require potentially biased, time-consuming and costly laboratory assays. The pivotal role of the Boman's index, indicative of the virus's cell surface binding potential, is underscored, enhancing the precision of detecting antigenic transitions. NIAViD's efficacy is not only in identifying influenza isolates that belong to novel antigenic clusters, but also in pinpointing potential sites driving significant antigenic changes, without the reliance on explicit modelling of haemagglutinin inhibition titres. We believe this approach holds promise to augment existing surveillance networks, offering timely insights for the development of updated, effective influenza vaccines. Consequently, NIAViD, in conjunction with other resources, could be used to support surveillance efforts and inform the development of updated influenza vaccines.

Exploring stroke discourse on Twitter through content and network analysis among Indian users.

The study aimed to understand stroke-related Twitter conversations in India, focusing on topics, message sources, reach, and influential users to provide insights to stakeholders regarding community needs for knowledge, support, and interventions. Geo-tagged Twitter posts focusing on stroke originating from India and, spanning from November 7, 2022, to February 28, 2023, were systematically obtained via the Twitter application programming interface, using keywords and hashtags sourced through Symplur Signals. Preprocessing involved the removal of hashtags, stop words, and URLs. The Latent Dirichlet Allocation (LDA) topic model was used to identify recurring stroke-related topics, while influential users were identified through social network analysis. About half of the tweets about stroke in India were about seeking support and post-stroke bereavement sharing and had the highest reachability. Four out of 10 tweets were from the individual twitter users. Tweets on the topic risk factors, awareness and prevention (14.6%) constituted the least proportion, whereas the topic management, research, and promotion had the least retweet ratio. Twitter demonstrates significant potential as a platform for both disseminating and acquiring stroke-related information within the Indian context. The identified topics and understanding of the content of discussion offer valuable resources to public health professionals and organizations to develop targeted educational and engagement strategies for the relevant audience.

Building RadiologyNET: an unsupervised approach to annotating a large-scale multimodal medical database.

BACKGROUND: The use of machine learning in medical diagnosis and treatment has grown significantly in recent years with the development of computer-aided diagnosis systems, often based on annotated medical radiology images. However, the lack of large annotated image datasets remains a major obstacle, as the annotation process is time-consuming and costly. This study aims to overcome this challenge by proposing an automated method for annotating a large database of medical radiology images based on their semantic similarity. RESULTS: An automated, unsupervised approach is used to create a large annotated dataset of medical radiology images originating from the Clinical Hospital Centre Rijeka, Croatia. The pipeline is built by data-mining three different types of medical data: images, DICOM metadata and narrative diagnoses. The optimal feature extractors are then integrated into a multimodal representation, which is then clustered to create an automated pipeline for labelling a precursor dataset of 1,337,926 medical images into 50 clusters of visually similar images. The quality of the clusters is assessed by examining their homogeneity and mutual information, taking into account the anatomical region and modality representation. CONCLUSIONS: The results indicate that fusing the embeddings of all three data sources together provides the best results for the task of unsupervised clustering of large-scale medical data and leads to the most concise clusters. Hence, this work marks the initial step towards building a much larger and more fine-grained annotated dataset of medical radiology images.

Machine Learning Approaches for Stroke Risk Prediction: Findings from the Suita Study.

Stroke constitutes a significant public health concern due to its impact on mortality and morbidity. This study investigates the utility of machine learning algorithms in predicting stroke and identifying key risk factors using data from the Suita study, comprising 7389 participants and 53 variables. Initially, unsupervised k-prototype clustering categorized participants into risk clusters, while five supervised models including Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), and Light Gradient Boosted Machine (LightGBM) were employed to predict stroke outcomes. Stroke incidence disparities among identified risk clusters using the unsupervised k-prototype clustering method are substantial, according to the findings. Supervised learning, particularly RF, was a preferable option because of the higher levels of performance metrics. The Shapley Additive Explanations (SHAP) method identified age, systolic blood pressure, hypertension, estimated glomerular filtration rate, metabolic syndrome, and blood glucose level as key predictors of stroke, aligning with findings from the unsupervised clustering approach in high-risk groups. Additionally, previously unidentified risk factors such as elbow joint thickness, fructosamine, hemoglobin, and calcium level demonstrate potential for stroke prediction. In conclusion, machine learning facilitated accurate stroke risk predictions and highlighted potential biomarkers, offering a data-driven framework for risk assessment and biomarker discovery.

Development of a Novel Classification Approach for Cow Behavior Analysis Using Tracking Data and Unsupervised Machine Learning Techniques.

Global Positioning Systems (GPSs) can collect tracking data to remotely monitor livestock well-being and pasture use. Supervised machine learning requires behavioral observations of monitored animals to identify changes in behavior, which is labor-intensive. Our goal was to identify animal behaviors automatically without using human observations. We designed a novel framework using unsupervised learning techniques. The framework contains two steps. The first step segments cattle tracking data using state-of-the-art time series segmentation algorithms, and the second step groups segments into clusters and then labels the clusters. To evaluate the applicability of our proposed framework, we utilized GPS tracking data collected from five cows in a 1096 ha rangeland pasture. Cow movement pathways were grouped into six behavior clusters based on velocity (m/min) and distance from water. Again, using velocity, these six clusters were classified into walking, grazing, and resting behaviors. The mean velocity for predicted walking and grazing and resting behavior was 44, 13 and 2 min/min, respectively, which is similar to other research. Predicted diurnal behavior patterns showed two primary grazing bouts during early morning and evening, like in other studies. Our study demonstrates that the proposed two-step framework can use unlabeled GPS tracking data to predict cattle behavior without human observations.