Dataset Information

Real-World Integration of a Sepsis Deep Learning Technology Into Routine Clinical Care: Implementation Study.

ABSTRACT: BACKGROUND:Successful integrations of machine learning into routine clinical care are exceedingly rare, and barriers to its adoption are poorly characterized in the literature. OBJECTIVE:This study aims to report a quality improvement effort to integrate a deep learning sepsis detection and management platform, Sepsis Watch, into routine clinical care. METHODS:In 2016, a multidisciplinary team consisting of statisticians, data scientists, data engineers, and clinicians was assembled by the leadership of an academic health system to radically improve the detection and treatment of sepsis. This report of the quality improvement effort follows the learning health system framework to describe the problem assessment, design, development, implementation, and evaluation plan of Sepsis Watch. RESULTS:Sepsis Watch was successfully integrated into routine clinical care and reshaped how local machine learning projects are executed. Frontline clinical staff were highly engaged in the design and development of the workflow, machine learning model, and application. Novel machine learning methods were developed to detect sepsis early, and implementation of the model required robust infrastructure. Significant investment was required to align stakeholders, develop trusting relationships, define roles and responsibilities, and to train frontline staff, leading to the establishment of 3 partnerships with internal and external research groups to evaluate Sepsis Watch. CONCLUSIONS:Machine learning models are commonly developed to enhance clinical decision making, but successful integrations of machine learning into routine clinical care are rare. Although there is no playbook for integrating deep learning into clinical care, learnings from the Sepsis Watch integration can inform efforts to develop machine learning technologies at other health care delivery systems.

SUBMITTER: Sendak MP

PROVIDER: S-EPMC7391165 | biostudies-literature | 2020 Jul

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Real-World Integration of a Sepsis Deep Learning Technology Into Routine Clinical Care: Implementation Study.

Sendak Mark P MP Ratliff William W Sarro Dina D Alderton Elizabeth E Futoma Joseph J Gao Michael M Nichols Marshall M Revoir Mike M Yashar Faraz F Miller Corinne C Kester Kelly K Sandhu Sahil S Corey Kristin K Brajer Nathan N Tan Christelle C Lin Anthony A Brown Tres T Engelbosch Susan S Anstrom Kevin K Elish Madeleine Clare MC Heller Katherine K Donohoe Rebecca R Theiling Jason J Poon Eric E Balu Suresh S Bedoya Armando A O'Brien Cara C

JMIR medical informatics 20200715 7

<h4>Background</h4>Successful integrations of machine learning into routine clinical care are exceedingly rare, and barriers to its adoption are poorly characterized in the literature.<h4>Objective</h4>This study aims to report a quality improvement effort to integrate a deep learning sepsis detection and management platform, Sepsis Watch, into routine clinical care.<h4>Methods</h4>In 2016, a multidisciplinary team consisting of statisticians, data scientists, data engineers, and clinicians was ...[more]

PMID: 32673244

Similar Datasets

Project description:BackgroundTo realize the broader benefits of electronic patient-reported outcome measures (ePROMs) in routine care, we used the RE-AIM (Reach, Effectiveness, Adoption, Implementation, and Maintenance) framework to inform the translation of a clinically effective ePROM system (hereafter referred to as the PRM system) into practice. The study aimed to evaluate the processes and success of implementing the PRM system in the routine care of patients diagnosed with lung cancer.MethodA controlled before-and-after mixed-methods study was undertaken. Data sources included a self-report questionnaire and interviews with healthcare providers, electronic health record data for PRMs patients and historical controls, and field notes. Descriptive statistics, logistic regression modelling, negative binomial models, generalized estimating equations and repeated measures ANOVA were used to analyze quantitative data. Qualitative data was thematically analyzed.ResultsA total of 48/79 eligible people diagnosed with lung cancer completed 90 assessments during the 5-month implementation period (RE-AIM reach). Every assessment breached the pre-defined threshold and care coordinators reviewed and actioned 95.6% of breaches, resulting in 146 referrals to allied health services, most frequently for social work (25.3%), dietetics (18.5%), physiotherapy (18.5%) and occupational therapy (17.1%). PRMs patients had significantly fewer visits to the cancer assessment unit for problematic symptoms (M = 0.23 vs. M = 0.43; p = 0.035), and were significantly more likely to be offered referrals (71% vs. 29%, p < 0.0001) than historical controls (RE-AIM effect). The levels of 'organizational readiness for implementing change' (ORIC) did not show much differences between baseline and follow-up, though this was already high at baseline; but significantly more staff reported improved confidence when asking patients to complete assessments (64.7% at baseline vs. 88.2% at follow-up, p = 0.0046), and when describing the assessment tool to patients (64.7% at baseline vs. 76.47% at follow-up, p = 0.0018) (RE-AIM adoption). A total of 78 staff received PRM system training, and 95.6% of the PRM system alerts were actioned (RE-AIM implementation); and all lung cancer care coordinators were engaged with the PRM system beyond the end of the study period (RE-AIM maintenance).ConclusionThis study demonstrates the potential of the PRM system in enhancing the routine care of lung cancer patients, through leveraging the capabilities of automated web-based care options. Research has shown the clear benefits of using electronically collected patient-reported outcome measures (ePROMs) for cancer patients and health services. However, we need to better understand how to implement ePROMs as part of routine care. This study evaluated the processes and outcomes of implementing an ePROMs system in the routine care of patients diagnosed with lung cancer. Key findings included: (a) a majority of eligible patients completed the scheduled assessments; (b) patient concerns were identified in every assessment, and care coordinators reviewed and actioned almost all of these, including making significantly more referrals to allied health services; (c) patients completing assessments regularly were less likely to present to the cancer assessment unit with problematic symptoms, suggesting that ePROMs identified patient concerns early and this led to a timely response to concerns; (d) staff training and engagement was high, and staff reporting increased confidence when asking patients to complete assessments and when describing the assessment tool to patients at the end of the implementation period. This study shows that implementing ePROMs in routine care is feasible and can lead to improvements in patient care.

Project description:BACKGROUND:Despite their high prevalence and significant burden, mental disorders such as depression remain largely underdiagnosed and undertreated. OBJECTIVE:The aim of the Allillanchu Project was to design, develop, and test an intervention to promote early detection, opportune referral, and access to treatment of patients with mental disorders attending public primary health care (PHC) services in Lima, Peru. METHODS:The project had a multiphase design: formative study, development of intervention components, and implementation. The intervention combined three strategies: training of PHC providers (PHCPs), task shifting the detection and referral of mental disorders, and a mobile health (mHealth) component comprising a screening app followed by motivational and reminder short message service (SMS) to identify at-risk patients. The intervention was implemented by 22 PHCPs from five health centers, working in antenatal care, tuberculosis, chronic diseases, and HIV or AIDS services. RESULTS:Over a period of 9 weeks, from September 2015 to November 2015, 733 patients were screened by the 22 PHCPs during routine consultations, and 762 screening were completed in total. The chronic diseases (49.9%, 380/762) and antenatal care services (36.7%, 380/762) had the higher number of screenings. Time constraints and workload were the main barriers to implementing the screening, whereas the use of technology, training, and supervision of the PHCPs by the research team were identified as facilitators. Of the 733 patients, 21.7% (159/733) screened positively and were advised to seek specialized care. Out of the 159 patients with a positive screening result, 127 had a follow-up interview, 72.4% (92/127) reported seeking specialized care, and 55.1% (70/127) stated seeing a specialist. Both patients and PHCPs recognized the utility of the screening and identified some key challenges to its wider implementation. CONCLUSIONS:The use of a screening app supported by training and supervision is feasible and uncovers a high prevalence of unidentified psychological symptoms in primary care. To increase its sustainability and utility, this procedure can be incorporated into the routine practices of existing health care services, following tailoring to the resources and features of each service. The early detection of psychological symptoms by a PHCP within a regular consultation, followed by adequate advice and support, can lead to a significant percentage of patients accessing specialized care and reducing the treatment gap of mental disorders.

Project description:In many mechanistic medical, biological, physical, and engineered spatiotemporal dynamic models the numerical solution of partial differential equations (PDEs), especially for diffusion, fluid flow and mechanical relaxation, can make simulations impractically slow. Biological models of tissues and organs often require the simultaneous calculation of the spatial variation of concentration of dozens of diffusing chemical species. One clinical example where rapid calculation of a diffusing field is of use is the estimation of oxygen gradients in the retina, based on imaging of the retinal vasculature, to guide surgical interventions in diabetic retinopathy. Furthermore, the ability to predict blood perfusion and oxygenation may one day guide clinical interventions in diverse settings, i.e., from stent placement in treating heart disease to BOLD fMRI interpretation in evaluating cognitive function (Xie et al., 2019; Lee et al., 2020). Since the quasi-steady-state solutions required for fast-diffusing chemical species like oxygen are particularly computationally costly, we consider the use of a neural network to provide an approximate solution to the steady-state diffusion equation. Machine learning surrogates, neural networks trained to provide approximate solutions to such complicated numerical problems, can often provide speed-ups of several orders of magnitude compared to direct calculation. Surrogates of PDEs could enable use of larger and more detailed models than are possible with direct calculation and can make including such simulations in real-time or near-real time workflows practical. Creating a surrogate requires running the direct calculation tens of thousands of times to generate training data and then training the neural network, both of which are computationally expensive. Often the practical applications of such models require thousands to millions of replica simulations, for example for parameter identification and uncertainty quantification, each of which gains speed from surrogate use and rapidly recovers the up-front costs of surrogate generation. We use a Convolutional Neural Network to approximate the stationary solution to the diffusion equation in the case of two equal-diameter, circular, constant-value sources located at random positions in a two-dimensional square domain with absorbing boundary conditions. Such a configuration caricatures the chemical concentration field of a fast-diffusing species like oxygen in a tissue with two parallel blood vessels in a cross section perpendicular to the two blood vessels. To improve convergence during training, we apply a training approach that uses roll-back to reject stochastic changes to the network that increase the loss function. The trained neural network approximation is about 1000 times faster than the direct calculation for individual replicas. Because different applications will have different criteria for acceptable approximation accuracy, we discuss a variety of loss functions and accuracy estimators that can help select the best network for a particular application. We briefly discuss some of the issues we encountered with overfitting, mismapping of the field values and the geometrical conditions that lead to large absolute and relative errors in the approximate solution.

Dataset Information

Real-World Integration of a Sepsis Deep Learning Technology Into Routine Clinical Care: Implementation Study.

Publications

Real-World Integration of a Sepsis Deep Learning Technology Into Routine Clinical Care: Implementation Study.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets