EMR Data Research Articles

Use of a Novel Process Map and Dashboard to Understand Drivers of Treatment Chart Turnaround Time in a Large Academic Radiation Oncology Practice

<h3>Purpose/Objective(s)</h3> Timely delivery of radiation treatment charts is critical to the safe and effective care of patients. Understanding the drivers of treatment chart turnaround time is required to ensure consistency and identify opportunities for improvement. Therefore, we sought to create a novel process map and dashboard for analysis and visualization of the treatment planning intervals that comprise turnaround time. <h3>Materials/Methods</h3> A comprehensive process map of treatment chart flow from simulation to image fusion, tumor volume contouring, treatment planning, physics check (QA), and therapist chart check was created by a team of physicians, radiation therapists, dosimetrists, physicists, and data analysts. Standardized interval definitions were developed to describe the time between unique steps in the planning process, and matched with corresponding quality check list items in the radiation oncology EMR. Time intervals between process steps were calculated for all treatment charts in the system in an automated and continuous fashion. A Tableau dashboard was created to analyze and display the data, allowing for analysis by treatment planning technique, disease site, attending physician, and department site. Results were verified by serial manual comparison of Tableau data with EMR data, repeated over five datasets until the dashboard algorithm successfully matched the data in the EMR. The resulting data were analyzed to identify median turnaround time (mTAT) from Simulation to Therapist Chart Check. To assess the incidence of late charts, a definition of "on time" treatment chart delivery was created for normal charts (excluding same-day and next-day treatments), defined as delivery of the chart at least 4 business hours prior to scheduled treatment start. The percentage of "on time" treatment charts was calculated and compared with a predefined institutional goal of 95%. <h3>Results</h3> A total of 3666 charts were included from 1 Jun 2021 to 1 Feb 2022, with data from 45 physicians across 10 department sites. The mTAT for all treatment charts was 4.80 business days. mTAT increased with increasing complexity of treatment planning and with certain disease sites (see table). Tumor volume completion time consumed the greatest proportion of overall turnaround time, at a median of 1.06 days for all charts and a median of 2.05 days and 2.29 days for IMRT and SBRT charts, respectively. About 93% (3018/3242) of normal charts were delivered "on time", which fell short of the predefined institutional goal of 95%. <h3>Conclusion</h3> The novel process map and dashboard accurately captured and calculated treatment planning intervals from EMR data and provided a platform to analyze and visualize drivers of mTAT and "on-time" chart delivery. This analysis identified contouring time as a key area of opportunity in reducing mTAT.

Durable viral suppression among persons with HIV in the deep south: an observational study

ABSTRACT This study assessed predictors of stable HIV viral suppression in a racially diverse sample of persons living with HIV (PWH) in the southern US. A total of 700 PWH were recruited from one of four HIV clinics in Metro Atlanta, GA. Data were collected from September 2012 to July 2017, and HIV viral loads were retrieved from EMR for 18 months. The baseline visits and EMR data were used for current analyses. Durable viral suppression was categorized as 1. Remain suppressed, 2. Remain unsuppressed, and 3. Unstable suppression. The number of antiretroviral medications and age were significantly associated with durable viral suppression. Older age, fewer ART medications and availability of social support were positively associated with durable viral suppression over the 18-month observation period. Findings suggest that regimen complexity is potentially a better predictor of viral suppression than self-reported medication adherence. The need for consensus on the definition of durable viral suppression is also urged.

DHDIP: An interpretable model for hypertension and hyperlipidemia prediction based on EMR data

Background and objectiveTraditional hypertension and hyperlipidemia prediction models suffer from uneven modeling data sources, small sample sizes, and a lack of uniform standards for the index system, resulting in the model failing to fulfill clinical applications. To address this issue, this work will offer DHDIP, an interpretable hypertension and hyperlipidemia prediction model based on EMR data. MethodsFirst, we will select massive high-dimensional, unstructured EMR data as a unified modeling data source, and propose a pre-processing algorithm for EMR data to solve the problem that EMR data cannot be directly processed by machine learning algorithms. Second, a variety of mainstream models such as XGBoost, CatBoost, and RandomForest are selected for modeling, and the best adaptation algorithms are identified by performance comparison. Finally, the SHAP framework was introduced into the DHDIP model, thus identifying the main factors contributing to hypertension and hyperlipidemia, effectively enhancing the interpretability of the model. ResultsThe DHDIP model's MSE value is 0.0285, and its LOSS value is 0.0054, both of which are better than previous studies. ConclusionThe model balances performance and interpretability. Multi-objective learning allows for a more thorough analysis and prediction of the condition, which not only lowers the cost of disease prediction but also aids physicians in clinical diagnosis. In addition, the datasets and source code are available from this link: https://github.com/Xiaoyao-Jia/DHDIP

Development of a claims-based flare algorithm for systemic lupus erythematosus

Objective To develop a claims-based algorithm identifying systemic lupus erythematosus (SLE) flares using a linked claims-electronic medical record (EMR) dataset. Methods This study was a retrospective analysis of linked administrative claims and EMR data spanning 1 January 2003 to 31 March 2019. Included were adult SLE patients with at least 12 months of continuous enrollment in claims data, 12 months of clinical activity in EMR, and an absence of malignancies excluding basal and squamous cell carcinoma. Patient follow-up was divided into 30-day windows, and a proxy SLEDAI-2K score based on the EMR data was calculated for each 30-day period. A flare was defined as an increase of at least 4 from the baseline score. A series of potential flare predictor variables identified in claims were based on a combination of established variables from a previous algorithm, with the addition of other SLE-related indicators based on clinical input. Logistic regression models were built to predict monthly SLE flares. Results Inclusion criteria identified 2427 patients. Results from a logistic model with forward selection capping the number of variables at 10 performed well with a c-statistic of 0.76 and a Brier score of 0.07. The top five predictors were any inpatient admission (OR = 4.76), outpatient office visit (OR = 3.04), MRI (OR = 2.26), ER visit (OR = 2.25), and number of rheumatology visits (OR = 1.75); p < .01 for all. Conclusions The final algorithm shows promise in providing an alternative and more streamlined way for identifying likely flares in administrative claims data that will advance the study of SLE within the context of flares.

HSD92 Natural Language Processing (NLP) of Unstructured EMR Data to Describe Treatment Patterns and Achievement of Guideline Recommended LDL-C Targets of Canadian Patients with Ascvd and Hypercholesterolemia

To describe treatment patterns of patients with ASCVD and hypercholesterolemia in Canada using novel NLP technology. Canadian lipid guidelines (GL) were recently (2021) updated, hence we also sought to understand how the change in LDL-C threshold will impact the current treatment gap.

Detection of high-grade glioma clusters: Spatial case-control study using EMR data from 2010-2021.

e14016 Background: High-grade gliomas (HGGs), like Glioblastomas (GBMs), are the most common malignant intracranial tumors with dismal prognosis. Rural areas have higher HGG related mortality and lesser research advancements. The objective of this study was to identify zip codes with disproportionate risk of HGGs compared to other brain malignancies. Methods: Patients’ electronic health data for this case-control study was extracted from West Virginia University Hospital Systems, 2010-2021. HGG cases were defined using diagnoses codes ICD10-C71 and ICD9-191 (excluding low grade gliomas). The control group was defined by using diagnoses codes ICD10-D32, C79.3, C71 (excluding HGG cases) and ICD9-225.2, 225.4, 198.3, 191 (excluding HGG cases). Controls were exact matched four controls to every case based upon patient five-year age categories and gender. Clustering was assessed globally through difference in Ripley’s-K between cases and controls; and locally using scan statistics. Results: The locations of HGGs and other brain tumors were clustered in the northern and northeastern regions of WV. Global cluster analyses detected statistically significant dispersion of HGG cases compared to our control group. Local cluster analysis identified multi-zip code cluster of HGG patients in northeastern WV between 2010-2021. The average relative risk of HGG compared to other brain malignancies was 1.25, indicating a 25% increase in risk of HGGs compared to other brain cancers within this region. While the local cluster was not statistically significant, the zip code level estimates of relative risk within the cluster ranged from 0.00 to 5.01. This indicated that certain zip codes within the cluster had five times the risk of HGGs compared to non-HGG brain malignancies relative to zip codes outside of the cluster. Conclusions: Findings indicate community level disparities in risk of HGGs relative to other brain malignancies. Additionally, results suggest that HGG cases are seen across the state at a more dispersed pattern than for non-HGG related brain tumors. These overall spatial trends could potentially be attributable to differences in clinical practice, access to care, or social determinants of health throughout WV. Spatial multivariable modeling techniques are needed to identify the specific neighborhood level factors associated with disproportionate patterns of risk identified for WV HGG patients.

EPRO-based digital symptom monitoring in a community oncology practice to reduce emergency room and inpatient utilization.

1508 Background: We have previously reported on the successful implementation of an electronic patient-reported outcomes (ePRO)-based symptom monitoring tool in a community oncology practice. Basch et al reported that use of such a tool in the academic trial setting reduced ER visits and hospitalizations. We have examined the impact of the tool on ER and inpatient utilization in this real world patient population. Methods: Highlands Oncology Group (HOG) is a 21 physician oncology group located in Northwest Arkansas. Beginning in June 2020, HOG offered patients receiving parenteral cancer therapy enrollment onto Expain, an EMR-integrated ePRO system which enables remote symptom monitoring during therapy. EMR data were linked with the Arkansas State Health Alliance for Records Exchange (SHARE), the state’s Health Information Exchange (HIE), to obtain ER visits/hospitalization data. All patients at HOG treated between September 30, 2020 and November 30, 2021 were included in this analysis. Clinical and demographic characteristics were compared in patients who enrolled on Expain versus those who did not, and corresponding p-values were calculated using Mann-Whitney and Chi-square tests. Crude rates for ER visits / hospitalizations were calculated as the total number of events per total person-time. Results: There were 855 patients enrolled on the ePRO system. Concurrently, in the same practice, 1773 patients were treated but not enrolled. Reasons for non-enrollment included patient’s choice to not participate and patient not yet offered enrollment due to rolling enrollment. The non-ePRO cohort was slightly older (66.7 vs 63.3 yrs, p &lt;.001), more commonly male (47.3% vs 39.3%, p &lt;.001) and less likely to be White (85.3% vs 89.4%, p = 0.003). The cohorts were comparable with respect to cancer site distribution and included a diverse and representative distribution of common malignancies receiving systemic therapy in a community practice. The proportion of patients with metastatic disease was comparable (ePRO 52.9% vs non-ePRO 51.6%, p = 0.55). Health resource utilization rates were lower for patients in the ePRO cohort: ER visits: 1.72 vs 2.34 per 100 patient-months, rate ratio and 95% CI = 0.74 (0.60, 0.92), p-value = 0.005; hospitalizations: 4.76 vs 5.41 per 100 patient-months, rate ratio and 95% CI = 0.87 (0.77, 0.99), p-value = 0.04. Conclusions: Our findings confirm the substantial benefits of using an ePRO tool in reducing health care resource utilization, and extend the initial findings of previous publications in the academic, clinical trial setting to the real world setting. This observational data is subject to confounding factors and we are evaluating the robustness using various methods to address non-comparability of the cohorts. We are further examining the benefits in specific patient subsets and attempting to correlate these benefits with improved survival.

Real-world (RW) treatment patterns and comparative effectiveness of Bruton tyrosine kinase inhibitors (BTKi) in patients (pts) with mantle cell lymphoma (MCL).

e18727 Background: BTKi therapies, approved for relapsed or refractory (R/R) MCL, have not been comprehensively evaluated in RW populations. This study aimed to assess patient characteristics, treatment patterns and associated outcomes in RW BTKi-treated MCL pts. Methods: The retrospective multicenter chart review was conducted in the Cardinal Health Oncology Provider Extended Network. EMR data were extracted for eligible pts diagnosed with MCL who initiated any of the approved BTKi (ibrutinib [ibr], acalabrutinib [acal], zanubrutinib [zanu]) from 2018 to 2021; pts enrolled in trials were excluded. Index date was defined as the use of any of the BTKis. Pts were followed 12-mo pre-index for medical history, and from index to last follow-up or death. Descriptive analyses were conducted to assess demographic/clinical characteristics, MCL baseline features, BTKi treatment patterns, adverse events (AE), and response rates by BTKi. Multivariable logistic regression was performed to assess factors associated with response and AE. Results: The study cohort consisted of 300 MCL pts (59% male; 69% white); most (64%) pts were covered by Medicare, 34% had commercial insurance. BTKis were given mainly as monotherapy (93%) and in R/R setting (86%). Pts in zanu group were significantly older (n = 100, median age = 71, range = 50-91) than pts in ibr (n = 100, median age = 69, range = 39-87) and acal (n = 100, median age = 70, range = 51-86) groups. Significantly fewer pts in the zanu group had baseline Ann Arbor stage I-II (4%) than ibr (10%) or acal (13%), while more zanu pts had presence of B symptoms (67%) than ibr (44%) or acal (57%). Pts in the zanu group also had significantly less with ECOG of 3+ (4%) compared to ibr (8%) or acal (6%). At BTKi initiation, significantly more pts in zanu group (18%) had history of atrial fibrillation than ibr (1%) or acal (5%). Multivariable regression reported a significant association of age, gender, extranodal/splenic involvement, and timing of BTKi initiation with response and AE (Table). Conclusions: This study provides the first RW evidence on comparative effectiveness of ibr, acal, zanu in MCL pts. While pts treated with zanu were older and had more complex MCL baseline features at initiation, multivariable regression suggested a trend favoring zanu over ibr or acal for both response and AE. Frontline initiation of BTKi therapy was also associated with improved tolerability. Future RW studies are needed to discern long-term outcomes.[Table: see text]

943-P: Impact of Culturally Appropriate Diabetes Care amongst Asian Americans with Type 2 Diabetes during the COVID Pandemic

Even as patients with T2D have been shown to have increased COVID-related morbidity and mortality, little is known on the impact of COVID on appointment adherence. This report reviews the impact of culturally-relevant diabetes clinics on adherence to clinical appointments, (pre-and during COVID) . This is a retrospective observational study using EMR data of adult patients with T2D managed at Joslin Diabetes Center before the COVID pandemic (April 20to Dec. 2019) and during the COVID pandemic (April 2020 to Dec. 2020) . Demographic and clinical visit data were extracted from patients seen at least once in both time periods. Clinical appointment adherence and outcomes were compared to patients in the general adult diabetes clinic (GADC) versus the Asian American Diabetes Initiative clinic (AADI) . 297 adult patients with T2D were seen in the AADI clinic and 4,655 patients were seen in the GADC clinics. The AADI clinic catered to a significantly larger proportion of patients who were non-native English speaking (46.1% vs. 8.7%, p&amp;lt;0.001) and on Medicaid (8.4% vs. 4.6%, p&amp;lt;0.001) . Early in the pandemic, virtual appointments were the major source of kept visits, with in-person kept appointments adherence returning in late 2020 (Oct-Dec.) . While the AADI clinic saw no significant changes in kept appointment rates, the GADC clinic had significantly fewer kept appointments overall compared to a similar time frame pre-COVID (pre: 71.7% vs. during: 63.5%, p=0.03) . Both clinics had significantly higher no-show rates during the pandemic. Comparing clinical outcomes, while GADC clinic patients saw a decline in their eGFR during the pandemic (80 vs. 75; p-value=0.04) , the AADI clinic did not observe a significant change in their patient’s renal status (82 vs. 83; p-value=0.93) . Our study shows that culturally oriented clinics like the AADI can successfully allow lower-incomes and older adults to acclimatize to telehealth services while also ensuring ongoing clinical wellbeing Disclosure A.Adam: None. K.Lau: None. H.Shah: None. M.Yu: None. G.L.King: Advisory Panel; Medtronic, Research Support; Janssen Pharmaceuticals, Inc.

138-LB: Improving Type 2 Diabetes Medication Prescribing

Background: For people with type 2 diabetes, implementing best practice guidelines for diabetes and cardiovascular comorbidities is important to reduce the risks of complications. Two adult medical clinics used population health EMR data reports to increase medication prescribing for statins, ACE &amp; ARB, SGLT2 &amp; GLP-1. Methods: The clinics tracked measures and used data analytics to identify a cohort population and medication needs for 6241 patients. Table 1 reflects data from a one-year pilot. Patients with a diagnosis of type 2 diabetes, duration of disease for at least one year, ages 18-75, with at least one in-person or telehealth visit were included. Results: Three medication prescribing baselines were tracked to improve type 2 diabetes and heart disease management. Prescribing rates of SGLT2 and GLP1 increased by 2%, ACE-I and ARB use increased by 1% and statins rates remained at 73%. Conclusion: The clinics used data to identify the cohort and prescribe necessary medications. Partnering with clinical pharmacists helped to support patients, educate providers, and standardize medication prescribing workflows. Planning care so that the teams consistently identified appropriate medications for diabetes and cardiovascular disease management were clinical priorities. While the goal rate of medication prescribing has not been achieved yet, new reporting workflows are now incorporated, setting up the clinics for continued improvement. Disclosure N. Rioles: None. M. Desimone: None. E. J. Hess: None. C. V. Iwummadu: None. E. L. Ospelt: None. S. Rompicherla: None. M. Greenfield: None. S. Tsai: None. Funding Helmsley Charitable Trust

QOPI Clinical Informatics: A digital platform to enable real-time quality reporting, clinical decision support, and rapid learning.

e13545 Background: The ASCO Quality Oncology Practice Initiative (QOPI) was established in 2006 to facilitate quality measurement and continuous improvement of cancer care.1 ASCO and QOPI leadership encourage a continuous learning environment to develop new tools and methodologies that improve quality and maximize outcomes and value in oncology practice.2 Parallel advances in tumor molecular profiling, bioinformatics, and EMR data curation are adding new insights to support the real-world research of longitudinal patient journeys. Methods: Referencing the QOPI Certification Track 2021 Measures Summary, we employed data visualization techniques to create a precision oncology dashboard for breast cancer. This evaluation included 409 breast cancer patients for whom genomic testing was performed and presented at the Avera molecular tumor board. Patients were then matched to appropriate biomarker-based targeted therapies. Results: The dashboard identified and prioritized areas for enhanced data capture to inform strategies for care-coordination, research, and delivery of specialized care. Conclusions: An interactive, visual platform may help facilitate the integration of QOPI reporting into CancerLinQ, thereby fostering a real-time, bidirectional system for quality reporting, clinical decision support, and rapid learning.3,4 The next stage will (1) expand the tumor list; and (2) apply NLP methods to assess accuracy, outcomes and temporal reasoning to create interactive graphs for patient referral patterns and surgery decision drivers.

The initial outcome of deploying a mortality prediction tool at community oncology practices.

1521 Background: Hospice improves the quality of life and care for cancer patients and reduces the likelihood of unwanted death in the hospital. Advance Care Planning (ACP) allows physicians to proactively initiate hospice and end-of-life discussions with identified patients, promoting timely hospice care enrollment. We developed a machine learning (ML) model to predict 90-day mortality risk for patients with metastatic cancer. The tool was designed to enable earlier ACP discussions leading to increased hospice enrollment. This study assesses the ML tool usage on ACP documentation in a community oncology setting. Methods: Twelve practices across the nation were included in the study, all participating in the Oncology Care Model. Five practices implemented the ML tool during 10/26/2020-9/30/2021, with patients scored every two weeks to provide insights on mortality risk. Patients identified as high-risk were evaluated for ACP utilization, obtained from timely EMR data and historical claims. Seven practices did not implement the ML tool and served as the control for the study. Results: A total of 1,663 patients were predicted to have a high risk of mortality at the 12 practices during the timeframe. The median age was 74 years. 53% of patients were males, and 47% were females. ACP documentation varied among the practices. The range was 19.4%-55.8% among ML tool participating practices and 7.4%-31.0% among non-participating practices. The weighted mean of ACP utilization was 34.4% for participating practices and 14.0% for non-participating practices. Compared with non-participating practices, the ACP rate increased significantly by 2.5-fold for participating practices (p-value = 0.03, two-sided T-test). Conclusions: This initial outcome study showed improved ACP documentation after deploying a mortality prediction tool in a community oncology setting. We are currently working on propensity score matching and regression analysis to reduce the effect of confounding factors such as practices, patient demographics, diagnosis, and treatment. Future studies will evaluate the impact of mortality tool use on other outcomes, including hospice enrollment, emergency department visits, and hospital admission. Implementing the mortality prediction tool is an ongoing effort with more practices planned to adopt.

Evaluation of a three-part equity intervention for women of color with breast cancer.

TPS1129 Background: Women of color (WOC) with breast cancer miss and fail key points in care due to racial disparities in cancer services. Conscious and unconscious bias means that these women are not treated in a timely way or as rigorously. They are offered fewer options, and the patient navigation and education needed for them to self-advocate is ignored. Black women in particular have the highest breast cancer death rates. Care delivery differs even independent of such variables as literacy, income, and education. Separately and incrementally, findings to date have shown the potential oof patient navigation, equity assessment, and mobile support to reverse these disparities. In pilot investigations –– partnering with academic cancer centers –– The Chrysalis Initiative (a nonprofit patient advocacy and research organization) has validated the potential of combining these approaches in a three-part intervention. Based on preliminary work, it is hypothesized patients will: experience significantly greater adherence to the recommended continuum of care, without disruption or barriers; more often seek second opinions and additional supportive resources, and engage in clinical trials; demonstrate less co-morbidities, through more preventive measures and healthful lifestyle; suffer fewer interactions perceived to be influenced by racism; undergo less financial distress, with guidance on managing and planning costs; feel more confident and more optimistic in their outlook; and achieve better clinical results and lower costs. Methods: To determine and document the full impact, both quantitatively and qualitatively, of the experimental three-part intervention, the trial is: delivering navigation/coaching services to the study population, using counselors who are predominantly WOC who have experienced the challenges of breast cancer care. Surveys and interviews pre- and post-intervention will add to impact measures; providing the experimental group with the BC Navi App on both iOS and Android devices . Developed in partnership with InTouch, the app supports engagement and tracking. It provides a dashboard for care evaluation to supplement EMR data; conducting an equity assessment of breast cancer services of partnering clinical programs through use of focus groups with staff, patients, and community; procedural checks; and data collection from center EHR systems. The review audits 40 domains of care based on NCCI, NCI, ASCO, ACR, and other standards. The assessment team works closely and collaboratively with each cancer center’s clinical and administrative staff to reveal disparities and find consensus on ways to close gaps. This trial in progress is randomizing 200 subjects to the coaching/mobile app intervention arm and 100 to benefits of the equity assessment only, with comparison to nonWOC with breast cancer. The trial aims to disseminate the three-part intervention in easily reproducible form.

RIDAB: Electronic medical record-integrated real world data platform for predicting and summarizing interactions in biomedical research from heterogeneous data resources

Background and objective: With the advent of bioinformatics, biological databases have been constructed to computerize data. Biological systems can be described as interactions and relationships between elements constituting the systems, and they are organized in various biomedical open databases. These open databases have been used in approaches to predict functional interactions such as protein-protein interactions (PPI), drug-drug interactions (DDI) and disease-disease relationships (DDR). However, just combining interaction data has limited effectiveness in predicting the complex relationships occurring in a whole context. Each contributing source contains information on each element in a specific field of knowledge but there is a lack of inter-disciplinary insight in combining them.Methods: In this study, we propose the RWD Integrated platform for Discovering Associations in Biomedical research (RIDAB) to predict interactions between biomedical entities. RIDAB is established as a graph network to construct a platform that predicts the interactions of target entities. Biomedical open database is combined with EMRs each representing a biomedical network and a real-world data. To integrate databases from different domains to build the platform, mapping of the vocabularies was required. In addition, the appropriate structure of the network and the graph embedding method to be used were needed to be selected to fit the tasks.Results: The feasibility of the platform was evaluated using node similarity and link prediction for drug repositioning task, a commonly used task for biomedical network. In addition, we compared the US Food and Drug Administration (FDA)-approved repositioned drugs with the predicted result. By integrating EMR database with biomedical networks, the platform showed increased f1 score in predicting repositioned drugs, from 45.62% to 57.26%, compared to platforms based on biomedical networks alone.Conclusions: This study demonstrates that the elements of biomedical research findings can be reflected by integrating EMR data with open-source biomedical networks. In addition, showed the feasibility of using the established platform to represent the integration of biomedical networks and reflected the relationship between real world networks.

Value Analysis of Neoadjuvant Radiotherapy for Breast Cancer after Modified Radical Mastectomy Based on Data Mining.

Over the last two generations, there has been a surge of interest in nonmutilating treatment for women with early breast cancer. Neoadjuvant radiation therapy, which is progressively being provided to breast cancer patients, could be used to decrease tumor burden while also providing an ability to examine treatment response. This paper aims to explore the effects of the initiation time of radiotherapy after modified adjuvant radical mastectomy on the prognosis of breast cancer. The EMR data can be used to mine hidden rules, which are of great significance for treatment and prognosis analysis. In collaboration with breast cancer, the appropriate prediction model and visualization method are selected and a visual analysis system for breast cancer group and treatment plan based on electronic medical record is constructed. Patients with multiple dimensions are reduced and clustered to form patient groups. The differences of characteristics among patient groups are intuitively displayed by using Nightingale diagram, word cloud, and time axis visualization methods. The support vector machine (SVM) model is used to predict the treatment scheme. The radiotherapy time after modified radical surgery in the two groups was within 15 weeks (observation group) and 15 weeks (routine group), respectively. The incidence of complications, local recurrence rate, progression-free survival, and quality of life scores of patients in the routine group and observation group were compared. The total incidence of complications differed significantly between the observation and routine groups. The physical function, material function, psychological function, and social function of the observation group were significantly higher than the routine group (P < 0.05). Radiotherapy within 15 weeks after modified radical mastectomy for breast cancer can not only reduce the local recurrence rate but also prolong the progression-free survival of patients, and the incidence of complications will not increase, which will greatly help improve the quality of life of patients.

Hospitalized Children With Limited English Proficiency: The Strive to Improve Analgesia.

Hospitalized Children With Limited English Proficiency: The Strive to Improve Analgesia.

Refining electronic medical records representation in manifold subspace

BackgroundElectronic medical records (EMR) contain detailed information about patient health. Developing an effective representation model is of great significance for the downstream applications of EMR. However, processing data directly is difficult because EMR data has such characteristics as incompleteness, unstructure and redundancy. Therefore, preprocess of the original data is the key step of EMR data mining. The classic distributed word representations ignore the geometric feature of the word vectors for the representation of EMR data, which often underestimate the similarities between similar words and overestimate the similarities between distant words. This results in word similarity obtained from embedding models being inconsistent with human judgment and much valuable medical information being lost.ResultsIn this study, we propose a biomedical word embedding framework based on manifold subspace. Our proposed model first obtains the word vector representations of the EMR data, and then re-embeds the word vector in the manifold subspace. We develop an efficient optimization algorithm with neighborhood preserving embedding based on manifold optimization. To verify the algorithm presented in this study, we perform experiments on intrinsic evaluation and external classification tasks, and the experimental results demonstrate its advantages over other baseline methods.ConclusionsManifold learning subspace embedding can enhance the representation of distributed word representations in electronic medical record texts. Reduce the difficulty for researchers to process unstructured electronic medical record text data, which has certain biomedical research value.

Applying QI-focused SPIDER approach to safer deprescribing for geriatric patients: Results of the Toronto Feasibility Study.

Context: More than 25% of Canadian seniors are prescribed 10+ different medications each year. There is a direct association between more medications and persistent high care needs/costs for seniors. Effective and appropriate deprescribing for seniors in primary care is needed. Objective: To present results of a feasibility study of the Structured Process Informed by Data, Evidence and Research (SPIDER) project aiming at improving safer deprescribing for complex older patients in community-based primary care. Study Design: Single-arm mixed methods study in Toronto, Ontario. Evaluation included participant interviews, focus groups, field notes and quantitative EMR data. Setting: Primary care practices affiliated with the University of Toronto Practice-Based Research Network (UTOPIAN). Population Studied: Patients aged 65+ years taking 10+ different medications. Intervention: 1) QI-focused Learning Collaboratives (LCs); 2) practice coaching; and 3) EMR data for audit & feedback. Outcome Measures: feasibility across eight dimensions: acceptability, demand, implementation, adaptation, integration, practicality, and efficacy. Results: Demand: 33 physicians and 24 allied health professionals from ten UTOPIAN practices and one community health centre participated in the Toronto LC. Implementation: the LC included a full day initial workshop, two short webinars, and a half-day summative congress over nine months. Practices had a monthly call with their QI coach and quarterly data reports. Adaptation, integration, practicality: teams developed various deprescribing tools and processes that were integratable to local context and existing practices. Acceptability: Teams perceived access to coaching as a valuable element of the approach. The initial review and validation of patients identified in the data reports were deemed time-consuming, particularly for under-resourced practices with a large cohort of target patient population. Access to pharmacist services and in-house QI and data support was considered two critical enablers to the sustainability of the approach. Efficacy: reductions in PIP prevalence and prevalence of patients with at least one PIP were 3.6% (p=.4) and 1.4% (p=.5), respectively. Conclusions: The SPIDER approach appears to be feasible. Access to coaching support and pharmacist services may enhance sustainability.

Randomized controlled trial of an audit and feedback cycle to improve adherence to Choosing Wisely Canada recommendations.

Context: Choosing Wisely Canada (CWC) aims to engage healthcare providers to reduce unnecessary care. Reducing overuse of antimicrobials and antipsychotics in patients with dementia in primary care settings remains an ongoing challenge. Audit and Feedback (A&F) interventions can facilitate practice change but can be improved by understanding factors affecting the likelihood of uptake of the feedback. Objective: To assess the impact of A&F interventions to reduce unnecessary prescribing of antibiotics and antipsychotics use and evaluate whether practice changes remain stable over time. Design: Clustered randomized controlled trial. Dataset: The Manitoba Primary Care Research Network (MaPCReN), a practice-based network that includes a repository containing de-identified EMR data from over 288,000 Manitobans. Population: Primary care providers participating in MaPCReN were randomized to three A&F groups: 1) Information on the relevant CWC recommendations; 2) practice specific data along with general information regarding the CWC recommendations 3) No CWC information. Outcome Measures: Statistics demonstrating changes in target prescriptions. Multivariate regression assessed characteristics of providers with improved prescribing. Results: 182 primary care providers were evaluated, 86.3% decreased the number of target prescriptions. More providers decreased prescribing in group1 (88.9%) or group2 (94.0%) compared to the control group 3 (76.7%) (p-value 0.02). There was no statistically significant difference between practice specific feedback and generic CWC information. An average of 46 antibiotic medications per provider were prescribed for viral indications in 2014/2015. This dropped to 15 in 2016/17 and was 18 in 2018/19. An average of 3 antipsychotic medications were prescribed for patients with dementia in 2014/2015. This dropped 7% in 2016/17 and remained stable in 2018/19. There were no statistically significant provider characteristics when considering all providers with improved prescribing, but male, rural and fee for service providers were more likely to decrease prescribing greater than the mean. Conclusion: Both practice specific and generic A&F information sent directly to primary care providers by a trusted source reduced potentially unnecessary prescriptions. This supports ongoing engagement with primary care providers in practice-based research networks to improve care and promote sustained practice changes.

Use of PHQ-9 for monitoring patients with depression in integrated primary care practices.

The PHQ-9 is a validated, short self-report questionnaire designed to be used in primary care for diagnosis and management of depression. We wished to assess its use for monitoring symptoms in patients being treated for depression. To measure the use of the PHQ-9 in 12 months following an elevated score in adult patients in primary care practices. Retrospective cohort study with descriptive analysis of use of PHQ-9 for follow up; stratified by age category, gender, severity and chronicity of symptoms, and whether referred to integrated behavioral health. EMR data including dates and results of PHQ-9, patient demographics, and referral data. Adults aged 18 and older with a PHQ-9 score of 10 or higher during calendar year 2018. No one was excluded. Primary outcome measure was the number of PHQ-9's done in 12 months following the initial elevated score (index score). Secondary outcome measures are a set of quality measures for depression care related to monitoring with PHQ-9 at 6 and 12 months following index score. Among all patients, mean number of PHQ-9 following index score was 2.1. This was increased for those with chronic symptoms (defined as having an elevated score at least once in calendar year 2017. It was increased slightly as severity of index symptoms increased. It was increased substantially for those referred to integrated behavioral health clinicians in the practices. The PHQ-9 is underutilized as an instrument for monitoring patients who are being treated for depression in primary care. Measurement based care has a strong evidence base and interventions to improve utilization of the PHQ-9 for monitoring are likely to improve depression outcomes in primary care.