RESUMO
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.
RESUMO
In the U.S. inpatient payment system, the Diagnosis-Related Group (DRG) is pivotal, but its assignment process is inefficient. The study introduces DRG-LLaMA, an advanced large language model (LLM) fine-tuned on clinical notes to enhance DRGs assignment. Utilizing LLaMA as the foundational model and optimizing it through Low-Rank Adaptation (LoRA) on 236,192 MIMIC-IV discharge summaries, our DRG-LLaMA -7B model exhibited a noteworthy macro-averaged F1 score of 0.327, a top-1 prediction accuracy of 52.0%, and a macro-averaged Area Under the Curve (AUC) of 0.986, with a maximum input token length of 512. This model surpassed the performance of prior leading models in DRG prediction, showing a relative improvement of 40.3% and 35.7% in macro-averaged F1 score compared to ClinicalBERT and CAML, respectively. Applied to base DRG and complication or comorbidity (CC)/major complication or comorbidity (MCC) prediction, DRG-LLaMA achieved a top-1 prediction accuracy of 67.8% and 67.5%, respectively. Additionally, our findings indicate that DRG-LLaMA 's performance correlates with increased model parameters and input context lengths.
RESUMO
BACKGROUND: Appropriate use of available inpatient beds is an ongoing challenge for US hospitals. Historical capacity goals of 80% to 85% may no longer serve the intended purpose of maximizing the resources of space, staff, and equipment. Numerous variables affect the input, throughput, and output of a hospital. Some of these variables include patient demand, regulatory requirements, coordination of patient flow between various systems, coordination of processes such as bed management and patient transfers, and the diversity of departments (both inpatient and outpatient) in an organization. METHODS: Mayo Clinic Health System in the Southwest Minnesota region of the US, a community-based hospital system primarily serving patients in rural southwestern Minnesota and part of Iowa, consists of 2 postacute care and 3 critical access hospitals. Our inpatient bed usage rates had exceeded 85%, and patient transfers from the region to other hospitals in the state (including Mayo Clinic in Rochester, Minnesota) had increased. To address these quality gaps, we used a blend of Agile project management methodology, rapid Plan-Do-Study-Act cycles, and a proactive approach to patient placement in the medical-surgical units as a quality improvement initiative. RESULTS: During 2 trial periods of the initiative, the main hub hospital (Mayo Clinic Health System hospital in Mankato) and other hospitals in the region increased inpatient bed usage while reducing total out-of-region transfers. CONCLUSION: Our novel approach to proactively managing bed capacity in the hospital allowed the region's only tertiary medical center to increase capacity for more complex and acute cases by optimizing the use of historically underused partner hospital beds.
Assuntos
Pacientes Internados , População Rural , Humanos , Melhoria de Qualidade , Hospitais Rurais , Instituições de Assistência AmbulatorialRESUMO
Background: Acute blood pressure (BP) reduction is standard of care after acute intracerebral haemorrhage (ICH). More acute BP reduction is associated with acute kidney injury (AKI). It is not known if the choice of antihypertensive medications affects the risk of AKI. Methods: We analysed data from the ATACH-II clinical trial. AKI was defined by the Kidney Disease: Improving Global Outcomes criteria. We analysed antihypertensive medication from two sources. The first was a case report form that specified the use of labetalol, diltiazem, urapidil or other. We tested the hypothesis that the secondary medication was associated with AKI with χ2 test. Second, we tested the hypotheses the dosage of diltiazem was associated with AKI using Mann-Whitney U test. Results: AKI occurred in 109 of 1000 patients (10.9%). A higher proportion of patients with AKI received diltiazem after nicardipine (12 (29%) vs 21 (12%), p=0.03). The 95%ile (90%-99% ile) of administered diltiazem was 18 (0-130) mg in patients with AKI vs 0 (0-30) mg in patients without AKI (p=0.002). There was no apparent confounding by indication for diltiazem use. Conclusions: The use of diltiazem, and more diltiazem, was associated with AKI in patients with acute ICH.
RESUMO
OBJECTIVES: Vaccines are crucial components of pandemic responses. Over 12 billion coronavirus disease 2019 (COVID-19) vaccines were administered at the time of writing. However, public perceptions of vaccines have been complex. We integrated social media and surveillance data to unravel the evolving perceptions of COVID-19 vaccines. MATERIALS AND METHODS: Applying human-in-the-loop deep learning models, we analyzed sentiments towards COVID-19 vaccines in 11 211 672 tweets of 2 203 681 users from 2020 to 2022. The diverse sentiment patterns were juxtaposed against user demographics, public health surveillance data of over 180 countries, and worldwide event timelines. A subanalysis was performed targeting the subpopulation of pregnant people. Additional feature analyses based on user-generated content suggested possible sources of vaccine hesitancy. RESULTS: Our trained deep learning model demonstrated performances comparable to educated humans, yielding an accuracy of 0.92 in sentiment analysis against our manually curated dataset. Albeit fluctuations, sentiments were found more positive over time, followed by a subsequence upswing in population-level vaccine uptake. Distinguishable patterns were revealed among subgroups stratified by demographic variables. Encouraging news or events were detected surrounding positive sentiments crests. Sentiments in pregnancy-related tweets demonstrated a lagged pattern compared with the general population, with delayed vaccine uptake trends. Feature analysis detected hesitancies stemmed from clinical trial logics, risks and complications, and urgency of scientific evidence. DISCUSSION: Integrating social media and public health surveillance data, we associated the sentiments at individual level with observed populational-level vaccination patterns. By unraveling the distinctive patterns across subpopulations, the findings provided evidence-based strategies for improving vaccine promotion during pandemics.
Assuntos
COVID-19 , Mídias Sociais , Feminino , Gravidez , Humanos , Vacinas contra COVID-19 , Análise de Sentimentos , COVID-19/prevenção & controle , Pandemias , Vigilância em Saúde PúblicaRESUMO
OBJECTIVE: Clinical knowledge-enriched transformer models (eg, ClinicalBERT) have state-of-the-art results on clinical natural language processing (NLP) tasks. One of the core limitations of these transformer models is the substantial memory consumption due to their full self-attention mechanism, which leads to the performance degradation in long clinical texts. To overcome this, we propose to leverage long-sequence transformer models (eg, Longformer and BigBird), which extend the maximum input sequence length from 512 to 4096, to enhance the ability to model long-term dependencies in long clinical texts. MATERIALS AND METHODS: Inspired by the success of long-sequence transformer models and the fact that clinical notes are mostly long, we introduce 2 domain-enriched language models, Clinical-Longformer and Clinical-BigBird, which are pretrained on a large-scale clinical corpus. We evaluate both language models using 10 baseline tasks including named entity recognition, question answering, natural language inference, and document classification tasks. RESULTS: The results demonstrate that Clinical-Longformer and Clinical-BigBird consistently and significantly outperform ClinicalBERT and other short-sequence transformers in all 10 downstream tasks and achieve new state-of-the-art results. DISCUSSION: Our pretrained language models provide the bedrock for clinical NLP using long texts. We have made our source code available at https://github.com/luoyuanlab/Clinical-Longformer, and the pretrained models available for public download at: https://huggingface.co/yikuan8/Clinical-Longformer. CONCLUSION: This study demonstrates that clinical knowledge-enriched long-sequence transformers are able to learn long-term dependencies in long clinical text. Our methods can also inspire the development of other domain-enriched long-sequence transformers.
Assuntos
Idioma , Aprendizagem , Processamento de Linguagem NaturalRESUMO
STUDY OBJECTIVE: The Geriatric Emergency Department Innovations (GEDI) program is a nurse-based geriatric assessment and care coordination program that reduces preventable admissions for older adults. Unfortunately, only 5% of older adults receive GEDI care because of resource limitations. The objective of this study was to predict the likelihood of hospitalization accurately and consistently with and without GEDI care using machine learning models to better target patients for the GEDI program. METHODS: We performed a cross-sectional observational study of emergency department (ED) patients between 2010 and 2018. Using propensity-score matching, GEDI patients were matched to other older adult patients. Multiple models, including random forest, were used to predict hospital admission. Multiple second-layer models, including random forest, were then used to predict whether GEDI assessment would change predicted hospital admission. Final model performance was reported as the area under the curve using receiver operating characteristic models. RESULTS: We included 128,050 patients aged over 65 years. The random forest ED disposition model had an area under the curve of 0.774 (95% confidence interval [CI] 0.741 to 0.806). In the random forest GEDI change-in-disposition model, 24,876 (97.3%) ED visits were predicted to have no change in disposition with GEDI assessment, and 695 (2.7%) ED visits were predicted to have a change in disposition with GEDI assessment. CONCLUSION: Our machine learning models could predict who will likely be discharged with GEDI assessment with good accuracy and thus select a cohort appropriate for GEDI care. In addition, future implementation through integration into the electronic health record may assist in selecting patients to be prioritized for GEDI care.
Assuntos
Serviço Hospitalar de Emergência , Hospitalização , Idoso , Humanos , Estudos Transversais , Aprendizado de Máquina , Avaliação Geriátrica , HospitaisRESUMO
The coronavirus disease 2019 pandemic had deleterious effects on the healthcare systems around the world. To increase intensive care units (ICUs) bed capacities, multiple adaptations had to be made to increase surge capacity. In this editorial, we demonstrate the changes made by an ICU of a midwest community hospital in the United States. These changes included moving patients that used to be managed in the ICU to progressive care units, such as patients requiring non-invasive ventilation and high flow nasal cannula, ST-elevation myocardial infarction patients, and post-neurosurgery patients. Additionally, newer tactics were applied to the processes of assessing oxygen supply and demand, patient care rounds, and post-ICU monitoring.
RESUMO
BACKGROUND: Machine learning (ML) approaches have been broadly applied to the prediction of length of stay and mortality in hospitalized patients. ML may also reduce societal health burdens, assist in health resources planning and improve health outcomes. However, the fairness of these ML models across ethnoracial or socioeconomic subgroups is rarely assessed or discussed. In this study, we aim (1) to quantify the algorithmic bias of ML models when predicting the probability of long-term hospitalization or in-hospital mortality for different heart failure (HF) subpopulations, and (2) to propose a novel method that can improve the fairness of our models without compromising predictive power. METHODS: We built 5 ML classifiers to predict the composite outcome of hospitalization length-of-stay and in-hospital mortality for 210 368 HF patients extracted from the Get With The Guidelines-Heart Failure registry data set. We integrated 15 social determinants of health variables, including the Social Deprivation Index and the Area Deprivation Index, into the feature space of ML models based on patients' geographies to mitigate the algorithmic bias. RESULTS: The best-performing random forest model demonstrated modest predictive power but selectively underdiagnosed underserved subpopulations, for example, female, Black, and socioeconomically disadvantaged patients. The integration of social determinants of health variables can significantly improve fairness without compromising model performance. CONCLUSIONS: We quantified algorithmic bias against underserved subpopulations in the prediction of the composite outcome for HF patients. We provide a potential direction to reduce disparities of ML-based predictive models by integrating social determinants of health variables. We urge fellow researchers to strongly consider ML fairness when developing predictive models for HF patients.
Assuntos
Insuficiência Cardíaca , Humanos , Feminino , Insuficiência Cardíaca/diagnóstico , Insuficiência Cardíaca/terapia , Tempo de Internação , Determinantes Sociais da Saúde , Hospitalização , Mortalidade HospitalarRESUMO
Machine learning is frequently being leveraged to tackle problems in the health sector including utilization for clinical decision-support. Its use has historically been focused on single modal data. Attempts to improve prediction and mimic the multimodal nature of clinical expert decision-making has been met in the biomedical field of machine learning by fusing disparate data. This review was conducted to summarize the current studies in this field and identify topics ripe for future research. We conducted this review in accordance with the PRISMA extension for Scoping Reviews to characterize multi-modal data fusion in health. Search strings were established and used in databases: PubMed, Google Scholar, and IEEEXplore from 2011 to 2021. A final set of 128 articles were included in the analysis. The most common health areas utilizing multi-modal methods were neurology and oncology. Early fusion was the most common data merging strategy. Notably, there was an improvement in predictive performance when using data fusion. Lacking from the papers were clear clinical deployment strategies, FDA-approval, and analysis of how using multimodal approaches from diverse sub-populations may improve biases and healthcare disparities. These findings provide a summary on multimodal data fusion as applied to health diagnosis/prognosis problems. Few papers compared the outputs of a multimodal approach with a unimodal prediction. However, those that did achieved an average increase of 6.4% in predictive accuracy. Multi-modal machine learning, while more robust in its estimations over unimodal methods, has drawbacks in its scalability and the time-consuming nature of information concatenation.
RESUMO
Despite the important role avoidable emergency department (ED) visits play in healthcare costs and quality of care, there has been little work in development of predictive models to identify patients likely to present with an avoidable ED visit. We use a conservative definition of 'avoidable' ED visits defined as visits that do not require diagnostic or screening services, procedures, or medications, and were discharged home to classify visits as avoidable. Models trained using data from emergency departments across the US yielded a training AUC of 0.723 and a testing AUC of 0.703. Models trained using the full dataset were tested on demographic groups (race, gender, insurance status), finding comparable performance between white/black patients and male/female with reductions in performance in Hispanic populations and patients with Medicaid. Predictors strongly associated with non-avoidable ED visits included increased age, increasing number of total chronic diseases, and general as well as digestive symptoms. Reasons for visit stated as injuries and psychiatric symptoms influenced the model to predict an avoidable visit.
RESUMO
BACKGROUND: Sepsis is one of the most life-threatening circumstances for critically ill patients in the United States, while diagnosis of sepsis is challenging as a standardized criteria for sepsis identification is still under development. Disparities in social determinants of sepsis patients can interfere with the risk prediction performances using machine learning. METHODS: We analyzed a cohort of critical care patients from the Medical Information Mart for Intensive Care (MIMIC)-III database. Disparities in social determinants, including race, sex, marital status, insurance types and languages, among patients identified by six available sepsis criteria were revealed by forest plots with 95% confidence intervals. Sepsis patients were then identified by the Sepsis-3 criteria. Sixteen machine learning classifiers were trained to predict in-hospital mortality for sepsis patients on a training set constructed by random selection. The performance was measured by area under the receiver operating characteristic curve (AUC). The performance of the trained model was tested on the entire randomly conducted test set and each sub-population built based on each of the following social determinants: race, sex, marital status, insurance type, and language. The fluctuations in performances were further examined by permutation tests. RESULTS: We analyzed a total of 11,791 critical care patients from the MIMIC-III database. Within the population identified by each sepsis identification method, significant differences were observed among sub-populations regarding race, marital status, insurance type, and language. On the 5783 sepsis patients identified by the Sepsis-3 criteria statistically significant performance decreases for mortality prediction were observed when applying the trained machine learning model on Asian and Hispanic patients, as well as the Spanish-speaking patients. With pairwise comparison, we detected performance discrepancies in mortality prediction between Asian and White patients, Asians and patients of other races, as well as English-speaking and Spanish-speaking patients. CONCLUSIONS: Disparities in proportions of patients identified by various sepsis criteria were detected among the different social determinant groups. The performances of mortality prediction for sepsis patients can be compromised when applying a universally trained model for each subpopulation. To achieve accurate diagnosis, a versatile diagnostic system for sepsis is needed to overcome the social determinant disparities of patients.
Assuntos
Sepse , Determinantes Sociais da Saúde , Estado Terminal , Mortalidade Hospitalar , Humanos , Aprendizado de Máquina , Estudos Retrospectivos , Sepse/diagnósticoRESUMO
Monitoring adverse drug events or pharmacovigilance has been promoted by the World Health Organization to assure the safety of medicines through a timely and reliable information exchange regarding drug safety issues. We aim to discuss the application of machine learning methods as well as causal inference paradigms in pharmacovigilance. We first reviewed data sources for pharmacovigilance. Then, we examined traditional causal inference paradigms, their applications in pharmacovigilance, and how machine learning methods and causal inference paradigms were integrated to enhance the performance of traditional causal inference paradigms. Finally, we summarized issues with currently mainstream correlation-based machine learning models and how the machine learning community has tried to address these issues by incorporating causal inference paradigms. Our literature search revealed that most existing data sources and tasks for pharmacovigilance were not designed for causal inference. Additionally, pharmacovigilance was lagging in adopting machine learning-causal inference integrated models. We highlight several currently trending directions or gaps to integrate causal inference with machine learning in pharmacovigilance research. Finally, our literature search revealed that the adoption of causal paradigms can mitigate known issues with machine learning models. We foresee that the pharmacovigilance domain can benefit from the progress in the machine learning field.
Assuntos
Efeitos Colaterais e Reações Adversas Relacionados a Medicamentos , Farmacovigilância , Causalidade , Efeitos Colaterais e Reações Adversas Relacionados a Medicamentos/epidemiologia , Humanos , Aprendizado de Máquina , Modelos TeóricosRESUMO
BACKGROUND: Seizures are a harmful complication of acute intracerebral hemorrhage (ICH). "Early" seizures in the first week after ICH are a risk factor for deterioration, later seizures, and herniation. Ideally, seizure medications after ICH would only be administered to patients with a high likelihood to have seizures. We developed and validated machine learning (ML) models to predict early seizures after ICH. METHODS: We used two large datasets to train and then validate our models in an entirely independent test set. The first model ("CAV") predicted early seizures from a subset of variables of the CAVE score (a prediction rule for later seizures)-cortical hematoma location, age less than 65 years, and hematoma volume greater than 10 mL-whereas early seizure was the dependent variable. We attempted to improve on the "CAV" model by adding anticoagulant use, antiplatelet use, Glasgow Coma Scale, international normalized ratio, and systolic blood pressure ("CAV + "). For each model we used logistic regression, lasso regression, support vector machines, boosted trees (Xgboost), and random forest models. Final model performance was reported as the area under the receiver operating characteristic curve (AUC) using receiver operating characteristic models for the test data. The setting of the study was two large academic institutions: institution 1, 634 patients; institution 2, 230 patients. There were no interventions. RESULTS: Early seizures were predicted across the ML models by the CAV score in test data, (AUC 0.72, 95% confidence interval 0.62-0.82). The ML model that predicted early seizure better in the test data was Xgboost (AUC 0.79, 95% confidence interval 0.71-0.87, p = 0.04) compared with the CAV model AUC. CONCLUSIONS: Early seizures after ICH are predictable. Models using cortical hematoma location, age less than 65 years, and hematoma volume greater than 10 mL had a good accuracy rate, and performance improved with more independent variables. Additional methods to predict seizures could improve patient selection for monitoring and prophylactic seizure medications.
Assuntos
Hemorragia Cerebral , Convulsões , Idoso , Hemorragia Cerebral/complicações , Escala de Coma de Glasgow , Hematoma/complicações , Humanos , Aprendizado de Máquina , Estudos Retrospectivos , Convulsões/diagnóstico , Convulsões/etiologiaRESUMO
Adrenal incidentalomas (AIs) are common incidental findings in medical practice with clinical significance. Although most AIs are nonsecretory and nonmalignant, they require a short course of follow-up over one to two years to rule out malignancy or hormonal secretion according to clinical practice guidelines. However, this can result in some adrenocortical carcinomas (ACCs) being missed if they transform at a later stage or evolve slowly. Here, we report one such case of an AI, which although remained indolent, eventually transformed into an ACC many years after the initial detection.