Unsupervised machine learning framework for discriminating major variants of concern during COVID-19.

Due to the high mutation rate of the virus, the COVID-19 pandemic evolved rapidly. Certain variants of the virus, such as Delta and Omicron emerged with altered viral properties leading to severe transmission and death rates. These variants burdened the medical systems worldwide with a major impact to travel, productivity, and the world economy. Unsupervised machine learning methods have the ability to compress, characterize, and visualize unlabelled data. This paper presents a framework that utilizes unsupervised machine learning methods to discriminate and visualize the associations between major COVID-19 variants based on their genome sequences. These methods comprise a combination of selected dimensionality reduction and clustering techniques. The framework processes the RNA sequences by performing a k-mer analysis on the data and further visualises and compares the results using selected dimensionality reduction methods that include principal component analysis (PCA), t-distributed stochastic neighbour embedding (t-SNE), and uniform manifold approximation projection (UMAP). Our framework also employs agglomerative hierarchical clustering to visualize the mutational differences among major variants of concern and country-wise mutational differences for selected variants (Delta and Omicron) using dendrograms. We also provide country-wise mutational differences for selected variants via dendrograms. We find that the proposed framework can effectively distinguish between the major variants and has the potential to identify emerging variants in the future.

Analysis of COVID-19 vaccine adverse event using language model and unsupervised machine learning.

BACKGROUND: After the COVID-19 pandemic, the world has made efforts to recover from the chaotic situation. Vaccination is a way to help control infectious diseases, and many people have been vaccinated against COVID-19 by this point. However, an extremely small number of those who received the vaccine have experienced diverse side effects. METHODS AND FINDINGS: In this study, we examined people who experienced adverse events with the COVID-19 vaccine by gender, age, vaccine manufacturer, and dose of vaccinations by using the Vaccine Adverse Event Reporting System datasets. Then we used a language model to vectorize symptom words and reduced their dimensionality. We also clustered symptoms by using unsupervised machine learning and analyzed the characteristics of each symptom cluster. Lastly, to discover any association rules among adverse events, we used a data mining approach. The frequency of adverse events was higher for women than men, for Moderna than for Pfizer or Janssen, and for the first dose than for the second dose. However, we found that characteristics of vaccine adverse events, including gender, vaccine manufacturer, age, and underlying diseases were different for each symptom cluster, and that fatal cases were significantly related to a particular cluster (associated with hypoxia). Also, as a result of the association analysis, the {chills ↔ pyrexia} and {vaccination site pruritus ↔ vaccination site erythema} rules had the highest support value of 0.087 and 0.046, respectively. CONCLUSIONS: We aim to contribute accurate information on the adverse events of the COVID-19 vaccine to relieve public anxiety due to unconfirmed statements about vaccines.

Evolution of Clinical Phenotypes of COVID-19 Patients During Intensive Care Treatment: An Unsupervised Machine Learning Analysis.

BACKGROUND: Identification of clinical phenotypes in critically ill COVID-19 patients could improve understanding of the disease heterogeneity and enable prognostic and predictive enrichment. However, previous attempts did not take into account temporal dynamics with high granularity. By including the dimension of time, we aim to gain further insights into the heterogeneity of COVID-19. METHODS: We used granular data from 3202 adult COVID patients in the Dutch Data Warehouse that were admitted to one of 25 Dutch ICUs between February 2020 and March 2021. Parameters including demographics, clinical observations, medications, laboratory values, vital signs, and data from life support devices were selected. Twenty-one datasets were created that each covered 24 h of ICU data for each day of ICU treatment. Clinical phenotypes in each dataset were identified by performing cluster analyses. Both evolution of the clinical phenotypes over time and patient allocation to these clusters over time were tracked. RESULTS: The final patient cohort consisted of 2438 COVID-19 patients with a ICU mortality outcome. Forty-one parameters were chosen for cluster analysis. On admission, both a mild and a severe clinical phenotype were found. After day 4, the severe phenotype split into an intermediate and a severe phenotype for 11 consecutive days. Heterogeneity between phenotypes appears to be driven by inflammation and dead space ventilation. During the 21-day period, only 8.2% and 4.6% of patients in the initial mild and severe clusters remained assigned to the same phenotype respectively. The clinical phenotype half-life was between 5 and 6 days for the mild and severe phenotypes, and about 3 days for the medium severe phenotype. CONCLUSIONS: Patients typically do not remain in the same cluster throughout intensive care treatment. This may have important implications for prognostic or predictive enrichment. Prominent dissimilarities between clinical phenotypes are predominantly driven by inflammation and dead space ventilation.

Rapidly identifying new coronavirus mutations of potential concern in the Omicron variant using an unsupervised learning strategy.

Extensive mutations in the Omicron spike protein appear to accelerate the transmission of SARS-CoV-2, and rapid infections increase the odds that additional mutants will emerge. To build an investigative framework, we have applied an unsupervised machine learning approach to 4296 Omicron viral genomes collected and deposited to GISAID as of December 14, 2021, and have identified a core haplotype of 28 polymutants (A67V, T95I, G339D, R346K, S371L, S373P, S375F, K417N, N440K, G446S, S477N, T478K, E484A, Q493R, G496S, Q498R, N501Y, Y505H, T547K, D614G, H655Y, N679K, P681H, N764K, K796Y, N856K, Q954H, N69K, L981F) in the spike protein and a separate core haplotype of 17 polymutants in non-spike genes: (K38, A1892) in nsp3, T492 in nsp4, (P132, V247, T280, S284) in 3C-like proteinase, I189 in nsp6, P323 in RNA-dependent RNA polymerase, I42 in Exonuclease, T9 in envelope protein, (D3, Q19, A63) in membrane glycoprotein, and (P13, R203, G204) in nucleocapsid phosphoprotein. Using these core haplotypes as reference, we have identified four newly emerging polymutants (R346, A701, I1081, N1192) in the spike protein (p value = 9.37*10-4, 1.0*10-15, 4.76*10-7 and 1.56*10-4, respectively), and five additional polymutants in non-spike genes (D343G in nucleocapsid phosphoprotein, V1069I in nsp3, V94A in nsp4, F694Y in the RNA-dependent RNA polymerase and L106L/F of ORF3a) that exhibit significant increasing trajectories (all p values < 1.0*10-15). In the absence of relevant clinical data for these newly emerging mutations, it is important to monitor them closely. Two emerging mutations may be of particular concern: the N1192S mutation in spike protein locates in an extremely highly conserved region of all human coronaviruses that is integral to the viral fusion process, and the F694Y mutation in the RNA polymerase may induce conformational changes that could impact remdesivir binding.

Clusters of Pregnant Women with Severe Acute Respiratory Syndrome Due to COVID-19: An Unsupervised Learning Approach.

COVID-19 has been widely explored in relation to its symptoms, outcomes, and risk profiles for the severe form of the disease. Our aim was to identify clusters of pregnant and postpartum women with severe acute respiratory syndrome (SARS) due to COVID-19 by analyzing data available in the Influenza Epidemiological Surveillance Information System of Brazil (SIVEP-Gripe) between March 2020 and August 2021. The study's population comprised 16,409 women aged between 10 and 49 years old. Multiple correspondence analyses were performed to summarize information from 28 variables related to symptoms, comorbidities, and hospital characteristics into a set of continuous principal components (PCs). The population was segmented into three clusters based on an agglomerative hierarchical cluster analysis applied to the first 10 PCs. Cluster 1 had a higher frequency of younger women without comorbidities and with flu-like symptoms; cluster 2 was represented by women who reported mainly ageusia and anosmia; cluster 3 grouped older women with the highest frequencies of comorbidities and poor outcomes. The defined clusters revealed different levels of disease severity, which can contribute to the initial risk assessment of the patient, assisting the referral of these women to health services with an appropriate level of complexity.

Public Discourse Surrounding Suicide during the COVID-19 Pandemic: An Unsupervised Machine Learning Analysis of Twitter Posts over a One-Year Period.

Many studies have forewarned the profound emotional and psychosocial impact of the protracted COVID-19 pandemic. This study thus aimed to examine how individuals relate to suicide amid the COVID-19 pandemic from a global perspective via the public Twitter discourse around suicide and COVID-19. Original Twitter tweets from 1 February 2020 to 10 February 2021 were searched, with terms related to "COVID-19", "suicide", or "self-harm". An unsupervised machine learning approach and topic modelling were used to identify topics from unique tweets, with each topic further grouped into themes using manually conducted thematic analysis by the study investigators. A total of 35,904 tweets related to suicide and COVID-19 were processed into 42 topics and six themes. The main themes were: (1) mixed reactions to COVID-19 public health policies and their presumed impact on suicide; (2) biopsychosocial impact of COVID-19 pandemic on suicide and self-harm; (3) comparing mortality rates of COVID-19, suicide, and other leading causes of death; (4) mental health support for individuals at risk of suicide; (5) reported cases and public reactions to news related to COVID-19, suicide, and homicide; and (6) figurative usage of the word suicide. The general public was generally concerned about governments' responses as well as the perturbing effects on mental health, suicide, the economy, and at-risk populations.

Unsupervised machine learning demonstrates the prognostic value of TAPSE/PASP ratio among hospitalized patients with COVID-19.

BACKGROUND: The ratio of tricuspid annular plane systolic excursion (TAPSE) to pulmonary artery systolic pressure (PASP) is a validated index of right ventricular-pulmonary arterial (RV-PA) coupling with prognostic value. We determined the predictive value of TAPSE/PASP ratio and adverse clinical outcomes in hospitalized patients with COVID-19. METHODS: Two hundred and twenty-nine consecutive hospitalized racially/ethnically diverse adults (≥18 years of age) admitted with COVID-19 between March and June 2020 with clinically indicated transthoracic echocardiograms (TTE) that included adequate tricuspid regurgitation (TR) velocities for calculation of PASP were studied. The exposure of interest was impaired RV-PA coupling as assessed by TAPSE/PASP ratio. The primary outcome was in-hospital mortality. Secondary endpoints comprised of ICU admission, incident acute respiratory distress syndrome (ARDS), and systolic heart failure. RESULTS: One hundred and seventy-six patients had both technically adequate TAPSE measurements and measurable TR velocities for analysis. After adjustment for age, sex, BMI, race/ethnicity, diabetes mellitus, and smoking status, log(TAPSE/PASP) had a significantly inverse association with ICU admission (p = 0.015) and death (p = 0.038). ROC analysis showed the optimal cutoff for TAPSE/PASP for death was 0.51 mm mmHg-1 (AUC = 0.68). Unsupervised machine learning identified two groups of echocardiographic function. Of all echocardiographic measures included, TAPSE/PASP ratio was the most significant in predicting in-hospital mortality, further supporting its significance in this cohort. CONCLUSION: Impaired RV-PA coupling, assessed noninvasively via the TAPSE/PASP ratio, was predictive of need for ICU level care and in-hospital mortality in hospitalized patients with COVID-19 suggesting utility of TAPSE/PASP in identification of poor clinical outcomes in this population both by traditional statistical and unsupervised machine learning based methods.

Variant-driven early warning via unsupervised machine learning analysis of spike protein mutations for COVID-19.

Never before such a vast amount of data, including genome sequencing, has been collected for any viral pandemic than for the current case of COVID-19. This offers the possibility to trace the virus evolution and to assess the role mutations play in its spread within the population, in real time. To this end, we focused on the Spike protein for its central role in mediating viral outbreak and replication in host cells. Employing the Levenshtein distance on the Spike protein sequences, we designed a machine learning algorithm yielding a temporal clustering of the available dataset. From this, we were able to identify and define emerging persistent variants that are in agreement with known evidences. Our novel algorithm allowed us to define persistent variants as chains that remain stable over time and to highlight emerging variants of epidemiological interest as branching events that occur over time. Hence, we determined the relationship and temporal connection between variants of interest and the ensuing passage to dominance of the current variants of concern. Remarkably, the analysis and the relevant tools introduced in our work serve as an early warning for the emergence of new persistent variants once the associated cluster reaches 1% of the time-binned sequence data. We validated our approach and its effectiveness on the onset of the Alpha variant of concern. We further predict that the recently identified lineage AY.4.2 ('Delta plus') is causing a new emerging variant. Comparing our findings with the epidemiological data we demonstrated that each new wave is dominated by a new emerging variant, thus confirming the hypothesis of the existence of a strong correlation between the birth of variants and the pandemic multi-wave temporal pattern. The above allows us to introduce the epidemiology of variants that we described via the Mutation epidemiological Renormalisation Group framework.

AI-driven quantification of ground glass opacities in lungs of COVID-19 patients using 3D computed tomography imaging.

OBJECTIVES: Ground-glass opacity (GGO)-a hazy, gray appearing density on computed tomography (CT) of lungs-is one of the hallmark features of SARS-CoV-2 in COVID-19 patients. This AI-driven study is focused on segmentation, morphology, and distribution patterns of GGOs. METHOD: We use an AI-driven unsupervised machine learning approach called PointNet++ to detect and quantify GGOs in CT scans of COVID-19 patients and to assess the severity of the disease. We have conducted our study on the "MosMedData", which contains CT lung scans of 1110 patients with or without COVID-19 infections. We quantify the morphologies of GGOs using Minkowski tensors and compute the abnormality score of individual regions of segmented lung and GGOs. RESULTS: PointNet++ detects GGOs with the highest evaluation accuracy (98%), average class accuracy (95%), and intersection over union (92%) using only a fraction of 3D data. On average, the shapes of GGOs in the COVID-19 datasets deviate from sphericity by 15% and anisotropies in GGOs are dominated by dipole and hexapole components. These anisotropies may help to quantitatively delineate GGOs of COVID-19 from other lung diseases. CONCLUSION: The PointNet++ and the Minkowski tensor based morphological approach together with abnormality analysis will provide radiologists and clinicians with a valuable set of tools when interpreting CT lung scans of COVID-19 patients. Implementation would be particularly useful in countries severely devastated by COVID-19 such as India, where the number of cases has outstripped available resources creating delays or even breakdowns in patient care. This AI-driven approach synthesizes both the unique GGO distribution pattern and severity of the disease to allow for more efficient diagnosis, triaging and conservation of limited resources.

Latent COVID-19 Clusters in Patients with Opioid Misuse.

The goal of this paper is to apply unsupervised machine learning techniques in order to discover latent clusters in patients who have opioid misuse and also undergone COVID-19 testing. Target dataset has been constructed based on COVID-19 testing results at Mount Sinai Health System and opioid treatment program (OTP) information from New York State Office of Addiction Service and Support (OASAS). The dataset was preprocessed using factor analysis for mixed data (FAMD) method and then K-means algorithm along with elbow method were used to determine the number of optimal clusters. Four patient clusters were identified among which the fourth cluster constituted the maximum percentage of positive COVID-19 test results (20%). Compared to the other clusters, this cluster has the highest percentage of African Americans. This cluster has also the highest mortality rate (16.52%), hospitalization rate after receiving the COVID-19 test result (72.17%, use of ventilator (7.83%) and ICU admission rate (47.83%). In addition, this cluster has the highest percentage of patients with at least one chronic disease (99.13%) and age-adjusted comorbidity score more than 1 (83.48%). Longer participation in OTP was associated with the highest morbidity and mortality from COVID-19.

Tracking SARS-CoV-2 Spike Protein Mutations in the United States (January 2020-March 2021) Using a Statistical Learning Strategy.

The emergence and establishment of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants of interest (VOIs) and variants of concern (VOCs) highlight the importance of genomic surveillance. We propose a statistical learning strategy (SLS) for identifying and spatiotemporally tracking potentially relevant Spike protein mutations. We analyzed 167,893 Spike protein sequences from coronavirus disease 2019 (COVID-19) cases in the United States (excluding 21,391 sequences from VOI/VOC strains) deposited at GISAID from 19 January 2020 to 15 March 2021. Alignment against the reference Spike protein sequence led to the identification of viral residue variants (VRVs), i.e., residues harboring a substitution compared to the reference strain. Next, generalized additive models were applied to model VRV temporal dynamics and to identify VRVs with significant and substantial dynamics (false discovery rate q-value < 0.01; maximum VRV proportion >10% on at least one day). Unsupervised learning was then applied to hierarchically organize VRVs by spatiotemporal patterns and identify VRV-haplotypes. Finally, homology modeling was performed to gain insight into the potential impact of VRVs on Spike protein structure. We identified 90 VRVs, 71 of which had not previously been observed in a VOI/VOC, and 35 of which have emerged recently and are durably present. Our analysis identified 17 VRVs ~91 days earlier than their first corresponding VOI/VOC publication. Unsupervised learning revealed eight VRV-haplotypes of four VRVs or more, suggesting two emerging strains (B1.1.222 and B.1.234). Structural modeling supported a potential functional impact of the D1118H and L452R mutations. The SLS approach equally monitors all Spike residues over time, independently of existing phylogenic classifications, and is complementary to existing genomic surveillance methods.

Triage and monitoring of COVID-19 patients in intensive care using unsupervised machine learning.

BACKGROUND: We designed an algorithm to assess COVID-19 patients severity and dynamic intubation needs and predict their length of stay using the breathing frequency (BF) and oxygen saturation (SpO2) signals. METHODS: We recorded the BF and SpO2 signals for confirmed COVID-19 patients admitted to the ICU of a teaching hospital during both the first and subsequent outbreaks of the pandemic in France. An unsupervised machine-learning algorithm (the Gaussian mixture model) was applied to the patients' data for clustering. The algorithm's robustness was ensured by comparing its results against actual intubation rates. We predicted intubation rates using the algorithm every hour, thus conducting a severity evaluation. We designed a S24 severity score that represented the patient's severity over the previous 24 h; the validity of MS24, the maximum S24 score, was checked against rates of intubation risk and prolonged ICU stay. RESULTS: Our sample included 279 patients. . The unsupervised clustering had an accuracy rate of 87.8% for intubation recognition (AUC = 0.94, True Positive Rate 86.5%, true Negative Rate 90.9%). The S24 score of intubated patients was significantly higher than that of non-intubated patients at 48 h before intubation. The MS24 score allowed for the distinguishing between three severity levels with an increased risk of intubation: green (3.4%), orange (37%), and red (77%). A MS24 score over 40 was highly predictive of an ICU stay greater than 5 days at an accuracy rate of 81.0% (AUC = 0.87). CONCLUSIONS: Our algorithm uses simple signals and seems to efficiently visualize the patients' respiratory situations, meaning that it has the potential to assist staffs' in decision-making. Additionally, real-time computation is easy to implement.

Public Opinions and Concerns Regarding the Canadian Prime Minister's Daily COVID-19 Briefing: Longitudinal Study of YouTube Comments Using Machine Learning Techniques.

BACKGROUND: During the COVID-19 pandemic in Canada, Prime Minister Justin Trudeau provided updates on the novel coronavirus and the government's responses to the pandemic in his daily briefings from March 13 to May 22, 2020, delivered on the official Canadian Broadcasting Corporation (CBC) YouTube channel. OBJECTIVE: The aim of this study was to examine comments on Canadian Prime Minister Trudeau's COVID-19 daily briefings by YouTube users and track these comments to extract the changing dynamics of the opinions and concerns of the public over time. METHODS: We used machine learning techniques to longitudinally analyze a total of 46,732 English YouTube comments that were retrieved from 57 videos of Prime Minister Trudeau's COVID-19 daily briefings from March 13 to May 22, 2020. A natural language processing model, latent Dirichlet allocation, was used to choose salient topics among the sampled comments for each of the 57 videos. Thematic analysis was used to classify and summarize these salient topics into different prominent themes. RESULTS: We found 11 prominent themes, including strict border measures, public responses to Prime Minister Trudeau's policies, essential work and frontline workers, individuals' financial challenges, rental and mortgage subsidies, quarantine, government financial aid for enterprises and individuals, personal protective equipment, Canada and China's relationship, vaccines, and reopening. CONCLUSIONS: This study is the first to longitudinally investigate public discourse and concerns related to Prime Minister Trudeau's daily COVID-19 briefings in Canada. This study contributes to establishing a real-time feedback loop between the public and public health officials on social media. Hearing and reacting to real concerns from the public can enhance trust between the government and the public to prepare for future health emergencies.

Leveraging Unsupervised Machine Learning to Discover Patterns in Linguistic Health Summaries for Eldercare.

The Center for Eldercare and Rehabilitation Technology, at University of Missouri, has researched the use of smart, unobtrusive sensors for older adult residents' health monitoring and alerting in aging-in-place communities for many years. Sensors placed in the apartments of older adult residents generate a deluge of daily data that is automatically aggregated, analyzed, and summarized to aid in health awareness, clinical care, and research for healthy aging. When anomalies or concerning trends are detected within the data, the sensor information is converted into linguistic health messages using fuzzy computational techniques, so as to make it understandable to the clinicians. Sensor data are analyzed at the individual level, therefore, through this study we aim to discover various combinations of patterns of anomalies happening together and recurrently in the older adult's population using these text summaries. Leveraging various computational text data processing techniques, we are able to extract relevant analytical features from the health messages. These features are transformed into a transactional encoding, then processed with frequent pattern mining techniques for association rule discovery. At individual level analysis, resident ID 3027 was considered as an exemplar to describe the analysis. Seven combinations of anomalies/rules/associations were discovered in this resident, out of which rule group three showed an increased recurrence during the COVID lockdown of facility. At the population level, a total of 38 associations were discovered that highlight the health patterns, and we continue to explore the health conditions associated with them. Ultimately, our goal is to correlate the combinations of anomalies with certain health conditions, which can then be leveraged for predictive analytics and preventative care. This will improve the current clinical care systems for older adult residents in smart sensor, aging-in-place communities.

Unsupervised machine learning reveals key immune cell subsets in COVID-19, rhinovirus infection, and cancer therapy.

For an emerging disease like COVID-19, systems immunology tools may quickly identify and quantitatively characterize cells associated with disease progression or clinical response. With repeated sampling, immune monitoring creates a real-time portrait of the cells reacting to a novel virus before disease-specific knowledge and tools are established. However, single cell analysis tools can struggle to reveal rare cells that are under 0.1% of the population. Here, the machine learning workflow Tracking Responders EXpanding (T-REX) was created to identify changes in both rare and common cells across human immune monitoring settings. T-REX identified cells with highly similar phenotypes that localized to hotspots of significant change during rhinovirus and SARS-CoV-2 infections. Specialized MHCII tetramer reagents that mark rhinovirus-specific CD4+ cells were left out during analysis and then used to test whether T-REX identified biologically significant cells. T-REX identified rhinovirus-specific CD4+ T cells based on phenotypically homogeneous cells expanding by ≥95% following infection. T-REX successfully identified hotspots of virus-specific T cells by comparing infection (day 7) to either pre-infection (day 0) or post-infection (day 28) samples. Plotting the direction and degree of change for each individual donor provided a useful summary view and revealed patterns of immune system behavior across immune monitoring settings. For example, the magnitude and direction of change in some COVID-19 patients was comparable to blast crisis acute myeloid leukemia patients undergoing a complete response to chemotherapy. Other COVID-19 patients instead displayed an immune trajectory like that seen in rhinovirus infection or checkpoint inhibitor therapy for melanoma. The T-REX algorithm thus rapidly identifies and characterizes mechanistically significant cells and places emerging diseases into a systems immunology context for comparison to well-studied immune changes.

Subtyping CKD Patients by Consensus Clustering: The Chronic Renal Insufficiency Cohort (CRIC) Study.

BACKGROUND: CKD is a heterogeneous condition with multiple underlying causes, risk factors, and outcomes. Subtyping CKD with multidimensional patient data holds the key to precision medicine. Consensus clustering may reveal CKD subgroups with different risk profiles of adverse outcomes. METHODS: We used unsupervised consensus clustering on 72 baseline characteristics among 2696 participants in the prospective Chronic Renal Insufficiency Cohort (CRIC) study to identify novel CKD subgroups that best represent the data pattern. Calculation of the standardized difference of each parameter used the cutoff of ±0.3 to show subgroup features. CKD subgroup associations were examined with the clinical end points of kidney failure, the composite outcome of cardiovascular diseases, and death. RESULTS: The algorithm revealed three unique CKD subgroups that best represented patients' baseline characteristics. Patients with relatively favorable levels of bone density and cardiac and kidney function markers, with lower prevalence of diabetes and obesity, and who used fewer medications formed cluster 1 (n=1203). Patients with higher prevalence of diabetes and obesity and who used more medications formed cluster 2 (n=1098). Patients with less favorable levels of bone mineral density, poor cardiac and kidney function markers, and inflammation delineated cluster 3 (n=395). These three subgroups, when linked with future clinical end points, were associated with different risks of CKD progression, cardiovascular disease, and death. Furthermore, patient heterogeneity among predefined subgroups with similar baseline kidney function emerged. CONCLUSIONS: Consensus clustering synthesized the patterns of baseline clinical and laboratory measures and revealed distinct CKD subgroups, which were associated with markedly different risks of important clinical outcomes. Further examination of patient subgroups and associated biomarkers may provide next steps toward precision medicine.

Examining the Utility of Social Media in COVID-19 Vaccination: Unsupervised Learning of 672,133 Twitter Posts.

BACKGROUND: Although COVID-19 vaccines have recently become available, efforts in global mass vaccination can be hampered by the widespread issue of vaccine hesitancy. OBJECTIVE: The aim of this study was to use social media data to capture close-to-real-time public perspectives and sentiments regarding COVID-19 vaccines, with the intention to understand the key issues that have captured public attention, as well as the barriers and facilitators to successful COVID-19 vaccination. METHODS: Twitter was searched for tweets related to "COVID-19" and "vaccine" over an 11-week period after November 18, 2020, following a press release regarding the first effective vaccine. An unsupervised machine learning approach (ie, structural topic modeling) was used to identify topics from tweets, with each topic further grouped into themes using manually conducted thematic analysis as well as guided by the theoretical framework of the COM-B (capability, opportunity, and motivation components of behavior) model. Sentiment analysis of the tweets was also performed using the rule-based machine learning model VADER (Valence Aware Dictionary and Sentiment Reasoner). RESULTS: Tweets related to COVID-19 vaccines were posted by individuals around the world (N=672,133). Six overarching themes were identified: (1) emotional reactions related to COVID-19 vaccines (19.3%), (2) public concerns related to COVID-19 vaccines (19.6%), (3) discussions about news items related to COVID-19 vaccines (13.3%), (4) public health communications about COVID-19 vaccines (10.3%), (5) discussions about approaches to COVID-19 vaccination drives (17.1%), and (6) discussions about the distribution of COVID-19 vaccines (20.3%). Tweets with negative sentiments largely fell within the themes of emotional reactions and public concerns related to COVID-19 vaccines. Tweets related to facilitators of vaccination showed temporal variations over time, while tweets related to barriers remained largely constant throughout the study period. CONCLUSIONS: The findings from this study may facilitate the formulation of comprehensive strategies to improve COVID-19 vaccine uptake; they highlight the key processes that require attention in the planning of COVID-19 vaccination and provide feedback on evolving barriers and facilitators in ongoing vaccination drives to allow for further policy tweaks. The findings also illustrate three key roles of social media in COVID-19 vaccination, as follows: surveillance and monitoring, a communication platform, and evaluation of government responses.

Identifying Communities at Risk for COVID-19-Related Burden Across 500 US Cities and Within New York City: Unsupervised Learning of the Coprevalence of Health Indicators.

BACKGROUND: Although it is well-known that older individuals with certain comorbidities are at the highest risk for complications related to COVID-19 including hospitalization and death, we lack tools to identify communities at the highest risk with fine-grained spatial resolution. Information collected at a county level obscures local risk and complex interactions between clinical comorbidities, the built environment, population factors, and other social determinants of health. OBJECTIVE: This study aims to develop a COVID-19 community risk score that summarizes complex disease prevalence together with age and sex, and compares the score to different social determinants of health indicators and built environment measures derived from satellite images using deep learning. METHODS: We developed a robust COVID-19 community risk score (COVID-19 risk score) that summarizes the complex disease co-occurrences (using data for 2019) for individual census tracts with unsupervised learning, selected on the basis of their association with risk for COVID-19 complications such as death. We mapped the COVID-19 risk score to corresponding zip codes in New York City and associated the score with COVID-19-related death. We further modeled the variance of the COVID-19 risk score using satellite imagery and social determinants of health. RESULTS: Using 2019 chronic disease data, the COVID-19 risk score described 85% of the variation in the co-occurrence of 15 diseases and health behaviors that are risk factors for COVID-19 complications among ~28,000 census tract neighborhoods (median population size of tracts 4091). The COVID-19 risk score was associated with a 40% greater risk for COVID-19-related death across New York City (April and September 2020) for a 1 SD change in the score (risk ratio for 1 SD change in COVID-19 risk score 1.4; P<.001) at the zip code level. Satellite imagery coupled with social determinants of health explain nearly 90% of the variance in the COVID-19 risk score in the United States in census tracts (r2=0.87). CONCLUSIONS: The COVID-19 risk score localizes risk at the census tract level and was able to predict COVID-19-related mortality in New York City. The built environment explained significant variations in the score, suggesting risk models could be enhanced with satellite imagery.

Using Unsupervised Machine Learning to Identify Age- and Sex-Independent Severity Subgroups Among Patients with COVID-19: Observational Longitudinal Study.

BACKGROUND: Early detection and intervention are the key factors for improving outcomes in patients with COVID-19. OBJECTIVE: The objective of this observational longitudinal study was to identify nonoverlapping severity subgroups (ie, clusters) among patients with COVID-19, based exclusively on clinical data and standard laboratory tests obtained during patient assessment in the emergency department. METHODS: We applied unsupervised machine learning to a data set of 853 patients with COVID-19 from the HM group of hospitals (HM Hospitales) in Madrid, Spain. Age and sex were not considered while building the clusters, as these variables could introduce biases in machine learning algorithms and raise ethical implications or enable discrimination in triage protocols. RESULTS: From 850 clinical and laboratory variables, four tests-the serum levels of aspartate transaminase (AST), lactate dehydrogenase (LDH), C-reactive protein (CRP), and the number of neutrophils-were enough to segregate the entire patient pool into three separate clusters. Further, the percentage of monocytes and lymphocytes and the levels of alanine transaminase (ALT) distinguished cluster 3 patients from the other two clusters. The highest proportion of deceased patients; the highest levels of AST, ALT, LDH, and CRP; the highest number of neutrophils; and the lowest percentages of monocytes and lymphocytes characterized cluster 1. Cluster 2 included a lower proportion of deceased patients and intermediate levels of the previous laboratory tests. The lowest proportion of deceased patients; the lowest levels of AST, ALT, LDH, and CRP; the lowest number of neutrophils; and the highest percentages of monocytes and lymphocytes characterized cluster 3. CONCLUSIONS: A few standard laboratory tests, deemed available in all emergency departments, have shown good discriminative power for the characterization of severity subgroups among patients with COVID-19.