PLOS digital health最新文献

Automated data transmission from diagnostic instrument networks to a central database at the Ministries of Health has the potential of providing real-time quality data not only on diagnostic instrument performance, but also continuous disease surveillance and patient care. We aimed at sharing how a locally developed novel diagnostic connectivity solution channels actionable data from diagnostic instruments to the national dashboards for disease control in Uganda between May 2022 and May 2023. The diagnostic connectivity solution was successfully configured on a selected network of multiplexing diagnostic instruments at 260 sites in Uganda, providing a layered access of data. Of these, 909,674 test results were automatically collected from 269 "GeneXpert" machines, 5597 test results from 28 "Truenat" and >12,000 were from 3 digital x-ray devices to different stakeholder levels to ensure optimal use of data for their intended purpose. The government and relevant stakeholders are empowered with usable and actionable data from the diagnostic instruments. The successful implementation of the diagnostic connectivity solution depended on some key operational strategies namely; sustained internet connectivity and short message services, stakeholder engagement, a strong in-country laboratory coordination network, human resource capacity building, establishing a network for the diagnostic instruments, and integration with existing health data collection tools. Poor bandwidth at some locations was a major hindrance for the successful implementation of the connectivity solution. Maintaining stakeholder engagement at the clinical level is key for sustaining diagnostic data connectivity. The locally developed diagnostic connectivity solution as a digital health technology offers the chance to collect high-quality data on a number of parameters for disease control, including error analysis, thereby strengthening the quality of data from the networked diagnostic sites to relevant stakeholders.

Specific moments of lapse among smokers attempting to quit often lead to full relapse, which highlights a need for interventions that target lapses before they might occur, such as just-in-time adaptive interventions (JITAIs). To inform the decision points and tailoring variables of a lapse prevention JITAI, we trained and tested supervised machine learning algorithms that use Ecological Momentary Assessments (EMAs) and wearable sensor data of potential lapse triggers and lapse incidence. We aimed to identify a best-performing and feasible algorithm to take forwards in a JITAI. For 10 days, adult smokers attempting to quit were asked to complete 16 hourly EMAs/day assessing cravings, mood, activity, social context, physical context, and lapse incidence, and to wear a Fitbit Charge 4 during waking hours to passively collect data on steps and heart rate. A series of group-level supervised machine learning algorithms (e.g., Random Forest, XGBoost) were trained and tested, without and with the sensor data. Their ability to predict lapses for out-of-sample (i) observations and (ii) individuals were evaluated. Next, a series of individual-level and hybrid (i.e., group- and individual-level) algorithms were trained and tested. Participants (N = 38) responded to 6,124 EMAs (with 6.9% of responses reporting a lapse). Without sensor data, the best-performing group-level algorithm had an area under the receiver operating characteristic curve (AUC) of 0.899 (95% CI = 0.871-0.928). Its ability to classify lapses for out-of-sample individuals ranged from poor to excellent (AUCper person = 0.524-0.994; median AUC = 0.639). 15/38 participants had adequate data for individual-level algorithms to be constructed, with a median AUC of 0.855 (range: 0.451-1.000). Hybrid algorithms could be constructed for 25/38 participants, with a median AUC of 0.692 (range: 0.523 to 0.998). With sensor data, the best-performing group-level algorithm had an AUC of 0.952 (95% CI = 0.933-0.970). Its ability to classify lapses for out-of-sample individuals ranged from poor to excellent (AUCper person = 0.494-0.979; median AUC = 0.745). 11/30 participants had adequate data for individual-level algorithms to be constructed, with a median AUC of 0.983 (range: 0.549-1.000). Hybrid algorithms could be constructed for 20/30 participants, with a median AUC of 0.772 (range: 0.444 to 0.968). In conclusion, high-performing group-level lapse prediction algorithms without and with sensor data had variable performance when applied to out-of-sample individuals. Individual-level and hybrid algorithms could be constructed for a limited number of individuals but had improved performance, particularly when incorporating sensor data for participants with sufficient wear time. Feasibility constraints and the need to balance multiple success criteria in the JITAI development and implementation process are discussed.

Introduction: Common, consumer-grade biosensors mounted on fitness trackers and smartwatches can measure an array of biometrics that have potential utility in post-discharge medical monitoring, especially in remote/rural communities. The feasibility characteristics for wrist-worn biosensors are poorly described for post-COVID conditions and rural populations.

Methods: We prospectively recruited patients in rural communities who were enrolled in an at-home rehabilitation program for post-COVID conditions. They were asked to wear a FitBit Charge 2 device and biosensor parameters were analyzed [e.g. heart rate, sleep, and activity]. Electronic patient reported outcome measures [E-PROMS] for mental [bi-weekly] and physical [daily] symptoms were collected using SMS text or email [per patient preference]. Exit surveys and interviews evaluated the patient experience.

Results: Ten patients were observed for an average of 58 days and half [N = 5] were monitored for 8 weeks or more. Five patients [50%] had been hospitalized with COVID [mean stay = 41 days] and 4 [36%] had required mechanical ventilation. As baseline, patients had moderate to severe levels of anxiety, depression, and stress; fatigue and shortness of breath were the most prevalent physical symptoms. Four patients [40%] already owned a smartwatch. In total, 575 patient days of patient monitoring occurred across 10 patients. Biosensor data was usable for 91.3% of study hours and surveys were completed 82.1% and 78.7% of the time for physical and mental symptoms, respectively. Positive correlations were observed between stress and resting heart rate [r = 0.360, p<0.01], stress and daily steps [r = 0.335, p<0.01], and anxiety and daily steps [r = 0.289, p<0.01]. There was a trend toward negative correlation between sleep time and physical symptom burden [r = -0.211, p = 0.05]. Patients reported an overall positive experience and identified the potential for wearable devices to improve medical safety and access to care. Concerns around data privacy/security were infrequent.

Conclusions: We report excellent feasibility characteristics for wrist-worn biosensors and e-PROMS as a possible substrate for multi-modal disease tracking in post-COVID conditions. Adapting consumer-grade wearables for medical use and scalable remote patient monitoring holds great potential.

Objectives: We aimed to use the high fidelity urgent care patient data to model the factors that have led to the increased demand at our local pediatric urgent care centre.

Methods: The dataset for this retrospective cohort study was obtained from our local healthcare centre's national reporting data for pediatric urgent care visits from 2006 to 2022. Variables analyzed included: basic patient demographics, chief complaint, triage urgency, date and time of registration/discharge, discharge diagnosis, and discharge destination. Statistical analysis of non-linear trends was summarized by locally estimated scatterplot smoothing splines. For machine learning, we used the tidymodels R package. Models were validated in training using k-fold cross validation with k = 5. We used univariate linear regression as a baseline model. After the data was standardized, correlation and homoscedasticity were evaluated between all parameter permutations.

Results: This dataset consisted of 164,660 unique visits to our academic centre's pediatric urgent care. Over the study period, there was an overall substantial increase in the number of urgent care visits per day, with a rapid increase beyond previous levels in 2021 and further in 2022. The increased length of stay trend was consistent across presenting complaint categories. The proportion of patients without primary care in 2022 was 2.5 times higher than in 2013. A random forest machine learning model revealed the relative importance of features to predicting a visit in 2022 were: longer stay, later registration in the day, diagnosis of an infectious illness, and younger age.

Conclusions: This study identified a combination of declining primary care access, circulating viral infections, and shifting chief complaints as factors driving the recent increase in frequency and duration of visits to our urgent care service.

Given the potential benefits of artificial intelligence and machine learning (AI/ML) within healthcare, it is critical to consider how these technologies can be deployed in pediatric research and practice. Currently, healthcare AI/ML has not yet adapted to the specific technical considerations related to pediatric data nor adequately addressed the specific vulnerabilities of children and young people (CYP) in relation to AI. While the greatest burden of disease in CYP is firmly concentrated in lower and middle-income countries (LMICs), existing applied pediatric AI/ML efforts are concentrated in a small number of high-income countries (HICs). In LMICs, use-cases remain primarily in the proof-of-concept stage. This narrative review identifies a number of intersecting challenges that pose barriers to effective AI/ML for CYP globally and explores the shifts needed to make progress across multiple domains. Child-specific technical considerations throughout the AI/ML lifecycle have been largely overlooked thus far, yet these can be critical to model effectiveness. Governance concerns are paramount, with suitable national and international frameworks and guidance required to enable the safe and responsible deployment of advanced technologies impacting the care of CYP and using their data. An ambitious vision for child health demands that the potential benefits of AI/Ml are realized universally through greater international collaboration, capacity building, strong oversight, and ultimately diffusing the AI/ML locus of power to empower researchers and clinicians globally. In order that AI/ML systems that do not exacerbate inequalities in pediatric care, teams researching and developing these technologies in LMICs must ensure that AI/ML research is inclusive of the needs and concerns of CYP and their caregivers. A broad, interdisciplinary, and human-centered approach to AI/ML is essential for developing tools for healthcare workers delivering care, such that the creation and deployment of ML is grounded in local systems, cultures, and clinical practice. Decisions to invest in developing and testing pediatric AI/ML in resource-constrained settings must always be part of a broader evaluation of the overall needs of a healthcare system, considering the critical building blocks underpinning effective, sustainable, and cost-efficient healthcare delivery for CYP.

With a renewed focus on health equity in the United States driven by national crises and legislation to improve digital healthcare innovation, there is a need for the designers of digital health tools to take deliberate steps to design for equity in their work. A concrete toolkit of methods to design for health equity is needed to support digital health practitioners in this aim. This narrative review summarizes several health equity frameworks to help digital health practitioners conceptualize the equity dimensions of importance for their work, and then provides design approaches that accommodate an equity focus. Specifically, the Double Diamond Model, the IDEAS framework and toolkit, and community collaboration techniques such as participatory design are explored as mechanisms for practitioners to solicit input from members of underserved groups and better design digital health tools that serve their needs. Each of these design methods requires a deliberate effort by practitioners to infuse health equity into the approach. A series of case studies that use different methods to build in equity considerations are offered to provide examples of how this can be accomplished and demonstrate the range of applications available depending on resources, budget, product maturity, and other factors. We conclude with a call for shared rigor around designing digital health tools that deliver equitable outcomes for members of underserved populations.

Contexts in which people drink vary. Certain drinking contexts may be more amenable to change than others and the effectiveness of alcohol reduction tactics may differ across contexts. This study aimed to explore how helpful context-specific tactics for alcohol reduction were perceived as being amongst increasing-and-higher-risk drinkers. Using the Behaviour Change Technique Taxonomy, context-specific tactics to reduce alcohol consumption were developed by the research team and revised following consultation with experts in behaviour change. In four focus groups (two online, two in-person), N = 20 adult increasing-and-higher-risk drinkers in the UK discussed how helpful tactics developed for four drinking contexts would be: drinking at home alone (19 tactics), drinking at home with partner or family (21 tactics), in the pub with friends (23 tactics), and a meal out of the home (20 tactics). Transcripts were analysed using constant comparison methods. Participants endorsed four broad approaches to reducing alcohol consumption which encompassed all the individual tactics developed by the research team: Diluting and substituting drinks for those containing less alcohol (e.g. switching to soft drinks or no- or low-alcohol drinks); Reducing external pressure to drink (e.g. setting expectations in advance); Creating barriers to drinking (e.g. not buying alcohol to keep at home or storing it in less visible places), and Setting new habits (e.g. breaking old patterns and taking up new hobbies). Three cross-cutting themes influenced how applicable these approaches were to different drinking contexts. These were: Situational pressure, Drinking motives, and Financial motivation. Diluting and substituting drinks which enabled covert reduction and Reducing external pressure to drink were favoured in social drinking contexts. Diluting and substituting drinks which enabled participants to feel that they were having 'a treat' or which facilitated relaxation and Creating barriers to drinking were preferred at home. Interventions to reduce alcohol consumption should offer tactics tailored to individuals' drinking contexts and which account for context-specific individual and situational pressure to drink.

Large language models (LLMs) have made a significant impact on the fields of general artificial intelligence. General purpose LLMs exhibit strong logic and reasoning skills and general world knowledge but can sometimes generate misleading results when prompted on specific subject areas. LLMs trained with domain-specific knowledge can reduce the generation of misleading information (i.e. hallucinations) and enhance the precision of LLMs in specialized contexts. Training new LLMs on specific corpora however can be resource intensive. Here we explored the use of a retrieval-augmented generation (RAG) model which we tested on literature specific to a biomedical research area. OpenAI's GPT-3.5, GPT-4, Microsoft's Prometheus, and a custom RAG model were used to answer 19 questions pertaining to diffuse large B-cell lymphoma (DLBCL) disease biology and treatment. Eight independent reviewers assessed LLM responses based on accuracy, relevance, and readability, rating responses on a 3-point scale for each category. These scores were then used to compare LLM performance. The performance of the LLMs varied across scoring categories. On accuracy and relevance, the RAG model outperformed other models with higher scores on average and the most top scores across questions. GPT-4 was more comparable to the RAG model on relevance versus accuracy. By the same measures, GPT-4 and GPT-3.5 had the highest scores for readability of answers when compared to the other LLMs. GPT-4 and 3.5 also had more answers with hallucinations than the other LLMs, due to non-existent references and inaccurate responses to clinical questions. Our findings suggest that an oncology research-focused RAG model may outperform general-purpose LLMs in accuracy and relevance when answering subject-related questions. This framework can be tailored to Q&A in other subject areas. Further research will help understand the impact of LLM architectures, RAG methodologies, and prompting techniques in answering questions across different subject areas.

This retrospective cross-sectional study investigates the impact of laboratory-specific hemoglobin reference intervals on electronic consultation (eConsult) referral patterns for suspected anemia and elevated hemoglobin at a tertiary care center in London, Ontario that serves Southwestern Ontario. The study analyzed referrals through the Ontario Telemedicine Network's eConsult platform for hemoglobin abnormalities, excluding patients under 18 years old, between July 1, 2019, and June 30, 2023.The main outcome measures were influence of hemoglobin reference intervals on the referral patterns for suspected anemia and elevated hemoglobin, as well as the extent of pre-referral laboratory testing. Of the 619 eConsults reviewed, 251 referrals for suspected anemia and 93 for elevated hemoglobin were analyzed. Referral patterns showed significant variance in hemoglobin levels based on different laboratory thresholds. Referrals for suspected anemia in females from laboratories whose lower limit was 120 g/L or greater had a hemoglobin concentration 7.5 g/L greater than referrals that used laboratories with a threshold lower than 120 g/L. The study also identified potential areas for improvement in pre-referral investigations; 44% of eConsults did not provide a ferritin level, 53% were missing a B12 level, and 81% were missing a reticulocyte count. In conclusion, laboratory reference intervals for hemoglobin significantly influence referral patterns for suspected hemoglobin abnormalities in Ontario's eConsult system. There is a need for standardized reference intervals and comprehensive pre-referral testing to avoid unnecessary medicalization and referrals. We propose an anemia management algorithm to guide primary care providers in the pre-referral investigation process.

It is expected but unknown whether machine-learning models can outperform regression models, such as a logistic regression (LR) model, especially when the number and types of predictor variables increase in electronic health records (EHRs). We aimed to compare the predictive performance of gradient-boosted decision tree (GBDT), random forest (RF), deep neural network (DNN), and LR with the least absolute shrinkage and selection operator (LR-LASSO) for unplanned readmission. We used EHRs of patients discharged alive from 38 hospitals in 2015-2017 for derivation and in 2018 for validation, including basic characteristics, diagnosis, surgery, procedure, and drug codes, and blood-test results. The outcome was 30-day unplanned readmission. We created six patterns of data tables having different numbers of binary variables (that ≥5% or ≥1% of patients or ≥10 patients had) with and without blood-test results. For each pattern of data tables, we used the derivation data to establish the machine-learning and LR models, and used the validation data to evaluate the performance of each model. The incidence of outcome was 6.8% (23,108/339,513 discharges) and 6.4% (7,507/118,074 discharges) in the derivation and validation datasets, respectively. For the first data table with the smallest number of variables (102 variables that ≥5% of patients had, without blood-test results), the c-statistic was highest for GBDT (0.740), followed by RF (0.734), LR-LASSO (0.720), and DNN (0.664). For the last data table with the largest number of variables (1543 variables that ≥10 patients had, including blood-test results), the c-statistic was highest for GBDT (0.764), followed by LR-LASSO (0.755), RF (0.751), and DNN (0.720), suggesting that the difference between GBDT and LR-LASSO was small and their 95% confidence intervals overlapped. In conclusion, GBDT generally outperformed LR-LASSO to predict unplanned readmission, but the difference of c-statistic became smaller as the number of variables was increased and blood-test results were used.