Traditional Machine Learning, Deep Learning, and BERT (Large Language Model) Approaches for Predicting Hospitalizations From Nurse Triage Notes: Comparative Evaluation of Resource Management.

BACKGROUND: Predicting hospitalization from nurse triage notes has the potential to augment care. However, there needs to be careful considerations for which models to choose for this goal. Specifically, health systems will have varying degrees of computational infrastructure available and budget constraints. OBJECTIVE: To this end, we compared the performance of the deep learning, Bidirectional Encoder Representations from Transformers (BERT)-based model, Bio-Clinical-BERT, with a bag-of-words (BOW) logistic regression (LR) model incorporating term frequency-inverse document frequency (TF-IDF). These choices represent different levels of computational requirements. METHODS: A retrospective analysis was conducted using data from 1,391,988 patients who visited emergency departments in the Mount Sinai Health System spanning from 2017 to 2022. The models were trained on 4 hospitals' data and externally validated on a fifth hospital's data. RESULTS: The Bio-Clinical-BERT model achieved higher areas under the receiver operating characteristic curve (0.82, 0.84, and 0.85) compared to the BOW-LR-TF-IDF model (0.81, 0.83, and 0.84) across training sets of 10,000; 100,000; and ~1,000,000 patients, respectively. Notably, both models proved effective at using triage notes for prediction, despite the modest performance gap. CONCLUSIONS: Our findings suggest that simpler machine learning models such as BOW-LR-TF-IDF could serve adequately in resource-limited settings. Given the potential implications for patient care and hospital resource management, further exploration of alternative models and techniques is warranted to enhance predictive performance in this critical domain. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): RR2-10.1101/2023.08.07.23293699.

Tele-medicine controlled hospital at home is associated with better outcomes than hospital stay.

BACKGROUND: Hospital-at-home (HAH) is increasingly becoming an alternative for in-hospital stay in selected clinical scenarios. Nevertheless, there is still a question whether HAH could be a viable option for acutely ill patients, otherwise hospitalized in departments of general-internal medicine. METHODS: This was a retrospective matched study, conducted at a telemedicine controlled HAH department, being part of a tertiary medical center. The objective was to compare clinical outcomes of acutely ill patients (both COVID-19 and non-COVID) admitted to either in-hospital or HAH. Non-COVID patients had one of three acute infectious diseases: urinary tract infections (UTI, either lower or upper), pneumonia, or cellulitis. RESULTS: The analysis involved 159 HAH patients (64 COVID-19 and 95 non-COVID) who were compared to a matched sample of in-hospital patients (192 COVID-19 and 285 non-COVID). The median length-of-hospital stay (LOS) was 2 days shorter in the HAH for both COVID-19 patients (95% CI: 1-3; p = 0.008) and non-COVID patients (95% CI; 1-3; p < 0.001). The readmission rates within 30 days were not significantly different for both COVID-19 patients (Odds Ratio (OR) = 1; 95% CI: 0.49-2.04; p = 1) and non-COVID patients (OR = 0.7; 95% CI; 0.39-1.28; p = 0.25). The differences remained insignificant within one year. The risk of death within 30 days was significantly lower in the HAH group for COVID-19 patients (OR = 0.34; 95% CI: 0.11-0.86; p = 0.018) and non-COVID patients (OR = 0.38; 95% CI: 0.14-0.9; p = 0.019). For one year survival period, the differences were significant for COVID-19 patients (OR = 0.5; 95% CI: 0.31-0.9; p = 0.044) and insignificant for non-COVID patients (OR = 0.63; 95% CI: 0.4-1; p = 0.052). CONCLUSIONS: Care for acutely ill patients in the setting of telemedicine-based hospital at home has the potential to reduce hospitalization length without increasing readmission risk and to reduce both 30 days and one-year mortality rates.

Advancing the future of equitable access to health care: recommendations from international health care leaders.

Disparities in access to health care are persistent and contribute to poor health outcomes for many populations around the world. Barriers to access are often similar across countries, despite differences in how health systems are structured. Health care leaders can work to address these barriers through bold, evidence-based actions. The Future of Health (FOH), an international community of senior health leaders, collaborated with the Duke-Margolis Institute for Health Policy to identify priority organizational and policy actions needed to improve equitable access to health care through a consensus-building exercise, a targeted literature review, and an expert discussion group. This paper describes four key action areas for health care leaders that FOH members identified as critical to enabling the future of equitable access to health care: ensuring prioritization of and accountability for equitable access to care; establishing comprehensive, organization-wide strategies to address barriers to access; clearly defining and incentivizing improvement on key measures related to reducing disparities in access; and establishing cross-sector partnerships to improve equitable access.

Advancing radiology practice and research: harnessing the potential of large language models amidst imperfections.

Large language models (LLMs) are transforming the field of natural language processing (NLP). These models offer opportunities for radiologists to make a meaningful impact in their field. NLP is a part of artificial intelligence (AI) that uses computer algorithms to study and understand text data. Recent advances in NLP include the Attention mechanism and the Transformer architecture. Transformer-based LLMs, such as GPT-4 and Gemini, are trained on massive amounts of data and generate human-like text. They are ideal for analysing large text data in academic research and clinical practice in radiology. Despite their promise, LLMs have limitations, including their dependency on the diversity and quality of their training data and the potential for false outputs. Albeit these limitations, the use of LLMs in radiology holds promise and is gaining momentum. By embracing the potential of LLMs, radiologists can gain valuable insights and improve the efficiency of their work. This can ultimately lead to improved patient care.

Machine learning in cardiac stress test interpretation: a systematic review.

Coronary artery disease (CAD) is a leading health challenge worldwide. Exercise stress testing is a foundational non-invasive diagnostic tool. Nonetheless, its variable accuracy prompts the exploration of more reliable methods. Recent advancements in machine learning (ML), including deep learning and natural language processing, have shown potential in refining the interpretation of stress testing data. Adhering to Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, we conducted a systematic review of ML applications in stress electrocardiogram (ECG) and stress echocardiography for CAD prognosis. Medical Literature Analysis and Retrieval System Online, Web of Science, and the Cochrane Library were used as databases. We analysed the ML models, outcomes, and performance metrics. Overall, seven relevant studies were identified. Machine-learning applications in stress ECGs resulted in sensitivity and specificity improvements. Some models achieved rates of above 96% in both metrics and reduced false positives by up to 21%. In stress echocardiography, ML models demonstrated an increase in diagnostic precision. Some models achieved specificity and sensitivity rates of up to 92.7 and 84.4%, respectively. Natural language processing applications enabled the categorization of stress echocardiography reports, with accuracy rates nearing 98%. Limitations include a small, retrospective study pool and the exclusion of nuclear stress testing, due to its well-documented status. This review indicates the potential of artificial intelligence applications in refining CAD stress testing assessment. Further development for real-world use is warranted.

Evaluating prompt engineering on GPT-3.5's performance in USMLE-style medical calculations and clinical scenarios generated by GPT-4.

This study was designed to assess how different prompt engineering techniques, specifically direct prompts, Chain of Thought (CoT), and a modified CoT approach, influence the ability of GPT-3.5 to answer clinical and calculation-based medical questions, particularly those styled like the USMLE Step 1 exams. To achieve this, we analyzed the responses of GPT-3.5 to two distinct sets of questions: a batch of 1000 questions generated by GPT-4, and another set comprising 95 real USMLE Step 1 questions. These questions spanned a range of medical calculations and clinical scenarios across various fields and difficulty levels. Our analysis revealed that there were no significant differences in the accuracy of GPT-3.5's responses when using direct prompts, CoT, or modified CoT methods. For instance, in the USMLE sample, the success rates were 61.7% for direct prompts, 62.8% for CoT, and 57.4% for modified CoT, with a p-value of 0.734. Similar trends were observed in the responses to GPT-4 generated questions, both clinical and calculation-based, with p-values above 0.05 indicating no significant difference between the prompt types. The conclusion drawn from this study is that the use of CoT prompt engineering does not significantly alter GPT-3.5's effectiveness in handling medical calculations or clinical scenario questions styled like those in USMLE exams. This finding is crucial as it suggests that performance of ChatGPT remains consistent regardless of whether a CoT technique is used instead of direct prompts. This consistency could be instrumental in simplifying the integration of AI tools like ChatGPT into medical education, enabling healthcare professionals to utilize these tools with ease, without the necessity for complex prompt engineering.

Artificial Intelligence Assessment of Biological Age From Transthoracic Echocardiography: Discrepancies with Chronologic Age Predict Significant Excess Mortality.

BACKGROUND: Age and sex can be estimated using artificial intelligence on the basis of various sources. The aims of this study were to test whether convolutional neural networks could be trained to estimate age and predict sex using standard transthoracic echocardiography and to evaluate the prognostic implications. METHODS: The algorithm was trained on 76,342 patients, validated in 22,825 patients, and tested in 20,960 patients. It was then externally validated using data from a different hospital (n = 556). Finally, a prospective cohort of handheld point-of-care ultrasound devices (n = 319; ClinicalTrials.gov identifier NCT05455541) was used to confirm the findings. A multivariate Cox regression model was used to investigate the association between age estimation and chronologic age with overall survival. RESULTS: The mean absolute error in age estimation was 4.9 years, with a Pearson correlation coefficient of 0.922. The probabilistic value of sex had an overall accuracy of 96.1% and an area under the curve of 0.993. External validation and prospective study cohorts yielded consistent results. Finally, survival analysis demonstrated that age prediction ≥5 years vs chronologic age was associated with an independent 34% increased risk for death during follow-up (P < .001). CONCLUSIONS: Applying artificial intelligence to standard transthoracic echocardiography allows the prediction of sex and the estimation of age. Machine-based estimation is an independent predictor of overall survival and, with further evaluation, can be used for risk stratification and estimation of biological age.

ChatGPT's adherence to otolaryngology clinical practice guidelines.

OBJECTIVES: Large language models, including ChatGPT, has the potential to transform the way we approach medical knowledge, yet accuracy in clinical topics is critical. Here we assessed ChatGPT's performance in adhering to the American Academy of Otolaryngology-Head and Neck Surgery guidelines. METHODS: We presented ChatGPT with 24 clinical otolaryngology questions based on the guidelines of the American Academy of Otolaryngology. This was done three times (N = 72) to test the model's consistency. Two otolaryngologists evaluated the responses for accuracy and relevance to the guidelines. Cohen's Kappa was used to measure evaluator agreement, and Cronbach's alpha assessed the consistency of ChatGPT's responses. RESULTS: The study revealed mixed results; 59.7% (43/72) of ChatGPT's responses were highly accurate, while only 2.8% (2/72) directly contradicted the guidelines. The model showed 100% accuracy in Head and Neck, but lower accuracy in Rhinology and Otology/Neurotology (66%), Laryngology (50%), and Pediatrics (8%). The model's responses were consistent in 17/24 (70.8%), with a Cronbach's alpha value of 0.87, indicating a reasonable consistency across tests. CONCLUSIONS: Using a guideline-based set of structured questions, ChatGPT demonstrates consistency but variable accuracy in otolaryngology. Its lower performance in some areas, especially Pediatrics, suggests that further rigorous evaluation is needed before considering real-world clinical use.

Home Versus Hospital Rehabilitation of Older Adults Following Hip Fracture Yields Similar Patient-Reported Outcome Measures.

The increase in hip fractures (HF) due to aging of the population and the rise in attractiveness of services provided at home following the COVID-19 pandemic, emphasize the need to compare outcomes of home versus hospital HF rehabilitation. To date, studies comparing the 2 services have focused primarily on clinical outcomes rather than patient-reported outcomes (PROs). This longitudinal observational study evaluated PROs of older adults with HF in the 2 settings. The SF36 questionnaire was used to measure PROs 3 times after surgery. The first PRO was retrospective and reflected pre-fracture health status. Descriptive statistics and mixed-effect logistic regression were used. Of 86 patients participating in the study, 41 had home rehabilitation and 45 had hospital rehabilitation. In both groups, the mental and physical scores plummeted 2 weeks after the HF, compared to pre-fracture status. The difference in improvement from pre-fracture status to recovery in both groups, were not significantly (P < .05) different, except for the pain domain. PROs of home versus hospital rehabilitation were similar, suggesting that rehabilitation at home can be as effective as hospital rehabilitation for suitable patients. This knowledge can improve quality of care in an aging global population.

Virtual reality's impact on children with type 1 diabetes: a proof-of-concept randomized cross-over trial on anxiety, pain, adherence, and glycemic control.

AIMS:  Assess the effectiveness of virtual reality (VR) technology, in reducing pain and anxiety, and improving adherence and glycemic control among children with type 1 diabetes (T1D). METHODS: Children with T1D, managed with continuous glucose monitoring and insulin pumps, were recruited for a randomized cross-over trial. Children were randomized to one of two interventions for diabetes management: group 1 used VR glasses first and group 2 listened to vocal-guided affective imagery first (audio). After 1 month, the interventions were crossed over. The outcome measures included pain and anxiety assessment, adherence, glycemic control, and patient-reported outcome measures (PROMs) of VR satisfaction and effectiveness. RESULTS:  Forty children, mean age 11.4 ± 1.8 years, were participated. During the VR part, the monthly mean pain score compared to the baseline improved in both groups by 30% (p = 0.03). A 14% reduction in the state anxiety score was observed from baseline to 1 month in both groups (p = 0.009). Glycemic control measures including time in range, time above range, and glucose management indicator improved in both groups during VR part (p < 0.004 for all), compared to audio part. After one month, the patient-reported outcome measure (PROM) of satisfaction and effectiveness was sixfold higher after 1 month in group 1 compared to group 2 (p = 0.002). Adherence improved for both groups. CONCLUSIONS: VR was shown to be effective in reducing pain and anxiety, improving adherence, PROM, and glycemic control among children with T1D. We suggest incorporating VR technology in pediatric diabetes clinics to facilitate and improve coping and management of diabetes. TRIAL REGISTRATION: Trial registration number and date of registration for prospectively registered trials:ClinicalTrials.gov Identifier: NCT05883267, May 10th, 2023.

Deep learning in voice analysis for diagnosing vocal cord pathologies: a systematic review.

OBJECTIVES: With smartphones and wearable devices becoming ubiquitous, they offer an opportunity for large-scale voice sampling. This systematic review explores the application of deep learning models for the automated analysis of voice samples to detect vocal cord pathologies. METHODS: We conducted a systematic literature review following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines. We searched MEDLINE and Embase databases for original publications on deep learning applications for diagnosing vocal cord pathologies between 2002 and 2022. Risk of bias was assessed using Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2). RESULTS: Out of the 14 studies that met the inclusion criteria, data from a total of 3037 patients were analyzed. All studies were retrospective. Deep learning applications targeted Reinke's edema, nodules, polyps, cysts, unilateral cord paralysis, and vocal fold cancer detection. Most pathologies had detection accuracy above 90%. Thirteen studies (93%) exhibited a high risk of bias and concerns about applicability. CONCLUSIONS: Technology holds promise for enhancing the screening and diagnosis of vocal cord pathologies. While current research is limited, the presented studies offer proof of concept for developing larger-scale solutions.

Sink-traps are a major source for carbapenemase-producing Enterobacteriaceae transmission.

OBJECTIVE: We studied the extent of carbapenemase-producing Enterobacteriaceae (CPE) sink contamination and transmission to patients in a nonoutbreak setting. METHODS: During 2017-2019, 592 patient-room sinks were sampled in 34 departments. Patient weekly rectal swab CPE surveillance was universally performed. Repeated sink sampling was conducted in 9 departments. Isolates from patients and sinks were characterized using pulsed-field gel electrophoresis (PFGE), and pairs of high resemblance were sequenced by Oxford Nanopore and Illumina. Hybrid assembly was used to fully assemble plasmids, which are shared between paired isolates. RESULTS: In total, 144 (24%) of 592 CPE-contaminated sinks were detected in 25 of 34 departments. Repeated sampling (n = 7,123) revealed that 52%-100% were contaminated at least once during the sampling period. Persistent contamination for >1 year by a dominant strain was common. During the study period, 318 patients acquired CPE. The most common species were Klebsiella pneumoniae, Escherichia coli, and Enterobacter spp. In 127 (40%) patients, a contaminated sink was the suspected source of CPE acquisition. For 20 cases with an identical sink-patient strain, temporal relation suggested sink-to-patient transmission. Hybrid assembly of specific sink-patient isolates revealed that shared plasmids were structurally identical, and SNP differences between shared pairs, along with signatures for potential recombination events, suggests recent sharing of the plasmids. CONCLUSIONS: CPE-contaminated sinks are an important source of transmission to patients. Although traditionally person-to-person transmission has been considered the main route of CPE transmission, these data suggest a change in paradigm that may influence strategies of preventing CPE dissemination.

Designing for flexibility in hybrid care services: lessons learned from a pilot in an internal medicine unit.

Digital transformation in healthcare during the COVID-19 pandemic led to the development of new hybrid models integrating physical and virtual care. The ability to provide remote care by telemedicine technologies and the need to better manage and control hospitals' occupancy accelerated growth in hospital-at-home programs. The Sheba Medical Center restructured to create Sheba Beyond as the first virtual hospital in Israel. These transformations enabled them to deliver hybrid services in their internal medicine unit by managing inpatient hospital-care with remote home-care based on the patients' medical condition. The hybrid services evolved to integrate care pathways multiplied by the mode of delivery-physical (in person) or virtual (technology enabled)-and the location of care-at the hospital or the patient home. The study examines this home hospitalization program pilot for internal medicine at Sheba Medical Center (MC). The research is based on qualitative semi-structured interviews with Sheba Beyond management, medical staff from the hospital and the Health Maintenance Organization (HMO), Architects, Information Technology (IT), Telemedicine and Medtech organizations. We investigated the implications of the development of hybrid services for the future design of the physical built-environment and the virtual technological platform. Our findings highlight the importance of designing for flexibility in the development of hybrid care services, while leveraging synergies across the built environment and digital platforms to support future models of care. In addition to exploring the potential for scalability in accelerating the flexibility of the healthcare system, we also highlight current barriers in professional, management, logistic and economic healthcare models.

Utilising patient-reported outcomes for goal-directed therapy of hip fracture patients: a sequential controlled trial.

BACKGROUND: Hip fracture patients (HFPs) frequently have multiple underlying conditions, necessitating that agreed-upon goals take these complications into consideration. Communication regarding goals between medical-personnel and patients is not always effective. Patient-reported outcomes (PROs) can outline personal goals and help promote quality health care in HFPs. Few studies have been published on this topic. The study's aim was to outline the process of using PROs for goal-directed therapy among HFPs. METHODS: This sequential controlled trial was conducted among HFPs from two medical centres. The control and the intervention group received integrative rehabilitation. PROs were measured in both groups using the SF36 questionnaire three times postsurgery: 24-48 hours, 2 weeks and 3 months. During the first round of questioning, only the intervention group was asked 'what matters most to you?' during the rehabilitative process. Accordingly, agreed-upon goals that were determined by the SF36's eight topics and were incorporated into the HFP's rehabilitative process. A Likert scale of 1-5, '1' indicating no-achievement and '5' full-achievement, was used to assess the goal achievement 4-6 months post-fracture. RESULTS: 84 HFPs participated in the study: 40 and 44 in the intervention and control group, respectively. In both groups, PROs declined after the HF, then improved somewhat 3 months later, but did not return to prefracture scores. Among the intervention group, 39% reached their specific goals (Likert level 5). Patients who achieved their goals had better PROs in comparison to others. The intervention group indicated PROs helped them articulate their desires and introduced them to new areas of care. CONCLUSIONS: Shifting from asking 'what's the matter?' to 'what matters most to you?' can improve the understanding of HFPs' own priorities, promote quality outcomes and enhance patient-centred care. Using PROs as a guide for goal-directed therapy can create a more inclusive process that includes the patients' most important health determinants and needs.

A Five-Decade Text Mining Analysis of Cochlear Implant Research: Where We Started and Where We Are Heading.

Background and Objectives: Since its invention in the 1970s, the cochlear implant (CI) has been substantially developed. We aimed to assess the trends in the published literature to characterize CI. Materials and Methods: We queried PubMed for all CI-related entries published during 1970-2022. The following data were extracted: year of publication, publishing journal, title, keywords, and abstract text. Search terms belonged to the patient's age group, etiology for hearing loss, indications for CI, and surgical methodological advancement. Annual trends of publications were plotted. The slopes of publication trends were calculated by fitting regression lines to the yearly number of publications. Results: Overall, 19,428 CIs articles were identified. Pediatric-related CI was the most dominant sub-population among the age groups, with the highest rate and slope during the years (slope 5.2 ± 0.3, p < 0.001), while elderly-related CIs had significantly fewer publications. Entries concerning hearing preservation showed the sharpest rise among the methods, from no entries in 1980 to 46 entries in 2021 (slope 1.7 ± 0.2, p < 0.001). Entries concerning robotic surgery emerged in 2000, with a sharp increase in recent years (slope 0.5 ± 0.1, p < 0.001). Drug-eluting electrodes and CI under local-anesthesia have been reported only in the past five years, with a gradual rise. Conclusions: Publications regarding CI among pediatrics outnumbered all other indications, supporting the rising, pivotal role of CI in the rehabilitation of children with sensorineural hearing loss. Hearing-preservation publications have recently rapidly risen, identified as the primary trend of the current era, followed by a sharp rise of robotic surgery that is evolving and could define the next revolution.

A blockchain-based computerized network infrastructure for the transparent, immutable calculation and dissemination of quantitative, measurable parameters of academic and medical research publications.

Digital transformation of healthcare systems should rely on decentralized computer networks and take advantage of the unique characteristics of blockchain technology. Decentralization ensures process transparency and data transparency for all relevant stakeholders. These values are essential in the realms of populations' healthcare information communications and processing, control and tracking of medical logistics supply chains, clinical research management, and control of certified healthcare services organizations. Mounting decentralized processes onto a blockchain-based computerized network will endow the values of immutability, improved cybersecurity, and potential for incentivizing stakeholders for relevant, pre-determined activities. One of the most relevant processes that would benefit from a decentralized, blockchain-based architecture is the submission, review, and publishing of scientific manuscripts. Current structures and processes in this world are non-transparent, poorly incentivizing significant stakeholders such as manuscripts' reviewers, and many are potentially corrupted. In this review, we suggest a blockchain-based architecture for such systems and advocate further research and development in several domains of modern healthcare systems-offering medicine to become "the new guy on the block (chain)."

Perception of Goals and Expected Outcomes in Older Hip Fracture Patients and Their Medical Staff: A Cross Sectional Study.

Background: Goal-oriented patientcare is a key element in qualityhealthcare. Medical-caregiver's (MC) are expected to generate a shared decision-making process with patients regarding goals and expected health-outcomes. Hip-fracture patients (HFP) are usually older-adults with multiple health-conditions, necessitating that agreed-upon goals regarding the rehabilitation process, take these conditions into consideration. This topic has yet to be investigated by pairing and comparing the perception of expected outcomes and therapeutic goals of multidisciplinary MCs and their HF patient's. Our aim was to assess in a quantitative method whether HFPs and their multidisciplinary MCs agree upon target health-outcomes and their most important goals as they are reflected in the SF12 questionnaire. Methods: This was a cross-sectional, multi-center, study of HFPs and their MCs. Patients and MCs were asked to rate their top three most important goals for rehabilitation from the SF12 eight subscales: physical functioning, physical role limitation, bodily pain, general health, vitality, social functioning, emotional role limitation and mental health, and indicate their expected outcome. Descriptive statistics and mixed effect logistic-regression were used to compare concordance of the ratings. Agreement between patients and MCs was assessed using interclass coefficients (ICCs). Results: A total of 378 ratings were collected from 52 patients, 12 nurses, 12 physicians and 6 paramedical personnel. Each patient had between 3 and 9 raters. Patients considered physical functioning and physical role limitation more important than did MCs. Physicians and nurses emphasized the importance of bodily pain while patients referred to it as relatively less significant. The total ICC was low (2%) indicating poor agreement between MCs and patients. With the exception of physical-functioning, MCs predicted a less optimistic outcome in all of the SF12's subscales in comparison to HFPs. Conclusion: Effective intervention in HFPs requires constructive communication between MCs and patients. The study suggests that caregivers have an insufficient understanding of the expectations of HFPs. More effective communication channels are required in order to better understand HFPs' needs and expectations.

Technologically assisted intensive home treatment: feasibility study.

Introduction: In recent year, many attempts have been made to provide patients with alternatives to psychiatric hospitalization during acute distress. Although several hospitalization alternatives have been offered, most of them still require patients to be distanced from their families, friends, and the social environment. Methods: In this report we describe the implementation of a novel approach to psychiatric care termed "Technologically assisted Intensive Home Treatment", where patients arriving to emergency settings are directed to home care with technological aids that enable close monitoring and ongoing contact with their therapists. Results: We describe the rationale and treatment principles of the treatment, and provide an elaborative description of the implementation process during the first year of implementation. Discussion: Additional attention is given to factors associated with early dropout from the program, in order to inform readers of predictors to optimal care. Limitations and directions for future research and practice are discussed.Clinical Trial Registration: The study was registered in the database of clinical trials (registration number SHEBA-19-6555-MW-CTIL) and in the Ministry of Health (registration number MOH_2022-08-22_011992).

Spatiotemporal analysis of small bowel capsule endoscopy videos for outcomes prediction in Crohn's disease.

Background: Deep learning techniques can accurately detect and grade inflammatory findings on images from capsule endoscopy (CE) in Crohn's disease (CD). However, the predictive utility of deep learning of CE in CD for disease outcomes has not been examined. Objectives: We aimed to develop a deep learning model that can predict the need for biological therapy based on complete CE videos of newly-diagnosed CD patients. Design: This was a retrospective cohort study. The study cohort included treatment-naïve CD patients that have performed CE (SB3, Medtronic) within 6 months of diagnosis. Complete small bowel videos were extracted using the RAPID Reader software. Methods: CE videos were scored using the Lewis score (LS). Clinical, endoscopic, and laboratory data were extracted from electronic medical records. Machine learning analysis was performed using the TimeSformer computer vision algorithm developed to capture spatiotemporal characteristics for video analysis. Results: The patient cohort included 101 patients. The median duration of follow-up was 902 (354-1626) days. Biological therapy was initiated by 37 (36.6%) out of 101 patients. TimeSformer algorithm achieved training and testing accuracy of 82% and 81%, respectively, with an Area under the ROC Curve (AUC) of 0.86 to predict the need for biological therapy. In comparison, the AUC for LS was 0.70 and for fecal calprotectin 0.74. Conclusion: Spatiotemporal analysis of complete CE videos of newly-diagnosed CD patients achieved accurate prediction of the need for biological therapy. The accuracy was superior to that of the human reader index or fecal calprotectin. Following future validation studies, this approach will allow for fast and accurate personalization of treatment decisions in CD.