Implications of the Use of Artificial Intelligence Predictive Models in Health Care Settings : A Simulation Study.

BACKGROUND: Substantial effort has been directed toward demonstrating uses of predictive models in health care. However, implementation of these models into clinical practice may influence patient outcomes, which in turn are captured in electronic health record data. As a result, deployed models may affect the predictive ability of current and future models. OBJECTIVE: To estimate changes in predictive model performance with use through 3 common scenarios: model retraining, sequentially implementing 1 model after another, and intervening in response to a model when 2 are simultaneously implemented. DESIGN: Simulation of model implementation and use in critical care settings at various levels of intervention effectiveness and clinician adherence. Models were either trained or retrained after simulated implementation. SETTING: Admissions to the intensive care unit (ICU) at Mount Sinai Health System (New York, New York) and Beth Israel Deaconess Medical Center (Boston, Massachusetts). PATIENTS: 130 000 critical care admissions across both health systems. INTERVENTION: Across 3 scenarios, interventions were simulated at varying levels of clinician adherence and effectiveness. MEASUREMENTS: Statistical measures of performance, including threshold-independent (area under the curve) and threshold-dependent measures. RESULTS: At fixed 90% sensitivity, in scenario 1 a mortality prediction model lost 9% to 39% specificity after retraining once and in scenario 2 a mortality prediction model lost 8% to 15% specificity when created after the implementation of an acute kidney injury (AKI) prediction model; in scenario 3, models for AKI and mortality prediction implemented simultaneously, each led to reduced effective accuracy of the other by 1% to 28%. LIMITATIONS: In real-world practice, the effectiveness of and adherence to model-based recommendations are rarely known in advance. Only binary classifiers for tabular ICU admissions data were simulated. CONCLUSION: In simulated ICU settings, a universally effective model-updating approach for maintaining model performance does not seem to exist. Model use may have to be recorded to maintain viability of predictive modeling. PRIMARY FUNDING SOURCE: National Center for Advancing Translational Sciences.

External Validation and Comparison of a General Ward Deterioration Index Between Diversely Different Health Systems.

OBJECTIVES: Implementing a predictive analytic model in a new clinical environment is fraught with challenges. Dataset shifts such as differences in clinical practice, new data acquisition devices, or changes in the electronic health record (EHR) implementation mean that the input data seen by a model can differ significantly from the data it was trained on. Validating models at multiple institutions is therefore critical. Here, using retrospective data, we demonstrate how Predicting Intensive Care Transfers and other UnfoReseen Events (PICTURE), a deterioration index developed at a single academic medical center, generalizes to a second institution with significantly different patient population. DESIGN: PICTURE is a deterioration index designed for the general ward, which uses structured EHR data such as laboratory values and vital signs. SETTING: The general wards of two large hospitals, one an academic medical center and the other a community hospital. SUBJECTS: The model has previously been trained and validated on a cohort of 165,018 general ward encounters from a large academic medical center. Here, we apply this model to 11,083 encounters from a separate community hospital. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: The hospitals were found to have significant differences in missingness rates (> 5% difference in 9/52 features), deterioration rate (4.5% vs 2.5%), and racial makeup (20% non-White vs 49% non-White). Despite these differences, PICTURE's performance was consistent (area under the receiver operating characteristic curve [AUROC], 0.870; 95% CI, 0.861-0.878), area under the precision-recall curve (AUPRC, 0.298; 95% CI, 0.275-0.320) at the first hospital; AUROC 0.875 (0.851-0.902), AUPRC 0.339 (0.281-0.398) at the second. AUPRC was standardized to a 2.5% event rate. PICTURE also outperformed both the Epic Deterioration Index and the National Early Warning Score at both institutions. CONCLUSIONS: Important differences were observed between the two institutions, including data availability and demographic makeup. PICTURE was able to identify general ward patients at risk of deterioration at both hospitals with consistent performance (AUROC and AUPRC) and compared favorably to existing metrics.

Preoperative Prostate MRI Predictors of Urinary Continence Following Radical Prostatectomy.

Background Urinary continence after radical prostatectomy (RP) is an important determinant of patient quality of life. Anatomic measures at prostate MRI have been previously associated with continence outcomes, but their predictive ability and interrater agreement are unclear in comprehensive clinical models. Purpose To evaluate the predictive ability and interrater agreement of MRI-based anatomic measurements of post-RP continence when combined with clinical multivariable models. Materials and Methods In this retrospective cohort study, continence outcomes were evaluated in men who underwent RP from August 2015 to October 2019. Preoperative MRI-based anatomic measures were obtained retrospectively by four abdominal radiologists. Before participation, these radiologists completed measure-specific training. Logistic regression models were developed with clinical variables alone, MRI variables alone, and combined variables for predicting continence at 3, 6, and 12 months after RP; some patient data were missing at each time point. Interrater agreement of MRI variables was assessed by using intraclass correlation coefficients (ICCs). Results A total of 586 men were included (mean age ± standard deviation: 63 years ± 7). The proportion of patients with incontinence was 0.2% (one of 589) at baseline, 27% (145 of 529) at 3 months, 14% (63 of 465) at 6 months, and 9% (37 of 425) at 12 months. Longer coronal membranous urethra length (MUL) improved the odds of post-RP continence at all time points (odds ratio per 1 mm: 0.86 [95% CI: 0.80, 0.93], P < .001; 0.86 [95% CI: 0.78, 0.95], P = .003; and 0.79 [95% CI: 0.67, 0.91], P = .002, respectively) in models that incorporated both clinical and MRI predictors. No other MRI variables were predictive. Age and baseline urinary function score were the only other predictive clinical variables at every time point. Interrater agreement was moderate (ICC, 0.62) for MUL among readers with measure-specific prostate MRI training and poor among those without the training (ICC, 0.38). Conclusion Preoperative MRI-measured coronal membranous urethra length was an independent predictor of urinary continence after prostatectomy. © RSNA, 2022 Online supplemental material is available for this article.

Development and Validation of Models to Predict Pathological Outcomes of Radical Prostatectomy in Regional and National Cohorts.

PURPOSE: Prediction models are recommended by national guidelines to support clinical decision making in prostate cancer. Existing models to predict pathological outcomes of radical prostatectomy (RP)-the Memorial Sloan Kettering (MSK) models, Partin tables, and the Briganti nomogram-have been developed using data from tertiary care centers and may not generalize well to other settings. MATERIALS AND METHODS: Data from a regional cohort (Michigan Urological Surgery Improvement Collaborative [MUSIC]) were used to develop models to predict extraprostatic extension (EPE), seminal vesicle invasion (SVI), lymph node invasion (LNI), and nonorgan-confined disease (NOCD) in patients undergoing RP. The MUSIC models were compared against the MSK models, Partin tables, and Briganti nomogram (for LNI) using data from a national cohort (Surveillance, Epidemiology, and End Results [SEER] registry). RESULTS: We identified 7,491 eligible patients in the SEER registry. The MUSIC model had good discrimination (SEER AUC EPE: 0.77; SVI: 0.80; LNI: 0.83; NOCD: 0.77) and was well calibrated. While the MSK models had similar discrimination to the MUSIC models (SEER AUC EPE: 0.76; SVI: 0.80; LNI: 0.84; NOCD: 0.76), they overestimated the risk of EPE, LNI, and NOCD. The Partin tables had inferior discrimination (SEER AUC EPE: 0.67; SVI: 0.76; LNI: 0.69; NOCD: 0.72) as compared to other models. The Briganti LNI nomogram had an AUC of 0.81 in SEER but overestimated the risk. CONCLUSIONS: New models developed using the MUSIC registry outperformed existing models and should be considered as potential replacements for the prediction of pathological outcomes in prostate cancer.

External Validation of Postpartum Hemorrhage Prediction Models Using Electronic Health Record Data.

OBJECTIVE: A recent study leveraging machine learning methods found that postpartum hemorrhage (PPH) can be predicted accurately at the time of labor admission in the U.S. Consortium for Safe Labor (CSL) dataset, with a C-statistic as high as 0.93. These CSL models were developed in older data (2002-2008) and used an estimated blood loss (EBL) of ≥1,000 mL to define PPH. We sought to externally validate these models using a more recent cohort of births where blood loss was measured using quantitative blood loss (QBL) methods. STUDY DESIGN: Using data from 5,261 deliveries between February 1, 2019 and May 11, 2020 at a single tertiary hospital, we mapped our electronic health record (EHR) data to the 55 predictors described in previously published CSL models. PPH was defined as QBL ≥1,000 mL within 24 hours after delivery. Model discrimination and calibration of the four CSL models were measured using our cohort. In a secondary analysis, we fit new models in our study cohort using the same predictors and algorithms as the original CSL models. RESULTS: The original study cohort had a substantially lower rate of PPH, 4.8% (7,279/228,438) versus 25% (1,321/5,261), possibly due to differences in measurement. The CSL models had lower discrimination in our study cohort, with a C-statistic as high as 0.57 (logistic regression). Models refit in our study cohort achieved better discrimination, with a C-statistic as high as 0.64 (random forest). Calibration improved in the refit models as compared with the original models. CONCLUSION: The CSL models' accuracy was lower in a contemporary EHR where PPH is assessed using QBL. As institutions continue to adopt QBL methods, further data are needed to understand the differences between EBL and QBL to enable accurate prediction of PPH. KEY POINTS: · Machine learning methods may help predict PPH.. · EBL models do not generalize when QBL is used.. · Blood loss estimation alters model accuracy..

Predictive Model-Driven Hotspotting to Decrease Emergency Department Visits: a Randomized Controlled Trial.

BACKGROUND: Emergency department (ED) visits contribute substantially to health care expenditures. Case management has been proposed as a strategy to address the medical and social needs of complex patients. However, strong research designs to evaluate the effectiveness of such interventions are limited. OBJECTIVES: To evaluate whether a community-based case management program was associated with reduced ED utilization among complex patients. DESIGN: Patients whose risk exceeded a threshold were randomly assigned to a group offered case management or to the control group. Assignment occurred at five intervals between November 2017 and January 2019. Program effectiveness for all assigned patients was assessed using an intention-to-treat effect. Program effectiveness among those who received treatment was assessed using a local average treatment effect, estimated using instrumental variables. Both estimators were adjusted for baseline characteristics using linear models. PARTICIPANTS: Adults over age 18 with at least one health care encounter with Michigan Medicine or St. Joseph Mercy Health System between June 2, 2016, and November 27, 2018. INTERVENTIONS: Intervention arm participants (n = 486) were offered coordinated case management across medical, mental health, and social service organizations. Control arm participants (n = 409) received usual care. MAIN MEASURES: The primary outcome was the number of ED visits in the 6 months following randomization into the study. Secondary outcomes were 6-month counts of inpatient and outpatient visits. KEY RESULTS: Of the 486 patients assigned to the intervention, 131 (27%) consented to receive case management. The intention-to-treat effect on ED visits was + 0.14 (95% CI: - 0.27 to + 0.55). The local average treatment effect among those who consented and received case management was + 0.53 (95% CI: - 1.00 to + 2.05). Intention-to-treat and local average treatment effects were not significant for secondary outcomes. CONCLUSIONS: The community case management intervention targeting ED visits was not associated with reduced utilization. Future case management interventions may benefit from additional patient engagement strategies and longer evaluation time periods. TRIAL REGISTRATION: Clinicaltrials.gov Identifier: NCT03293160.

COVID-19 Discourse on Twitter in Four Asian Countries: Case Study of Risk Communication.

BACKGROUND: COVID-19, caused by SARS-CoV-2, has led to a global pandemic. The World Health Organization has also declared an infodemic (ie, a plethora of information regarding COVID-19 containing both false and accurate information circulated on the internet). Hence, it has become critical to test the veracity of information shared online and analyze the evolution of discussed topics among citizens related to the pandemic. OBJECTIVE: This research analyzes the public discourse on COVID-19. It characterizes risk communication patterns in four Asian countries with outbreaks at varying degrees of severity: South Korea, Iran, Vietnam, and India. METHODS: We collected tweets on COVID-19 from four Asian countries in the early phase of the disease outbreak from January to March 2020. The data set was collected by relevant keywords in each language, as suggested by locals. We present a method to automatically extract a time-topic cohesive relationship in an unsupervised fashion based on natural language processing. The extracted topics were evaluated qualitatively based on their semantic meanings. RESULTS: This research found that each government's official phases of the epidemic were not well aligned with the degree of public attention represented by the daily tweet counts. Inspired by the issue-attention cycle theory, the presented natural language processing model can identify meaningful transition phases in the discussed topics among citizens. The analysis revealed an inverse relationship between the tweet count and topic diversity. CONCLUSIONS: This paper compares similarities and differences of pandemic-related social media discourse in Asian countries. We observed multiple prominent peaks in the daily tweet counts across all countries, indicating multiple issue-attention cycles. Our analysis identified which topics the public concentrated on; some of these topics were related to misinformation and hate speech. These findings and the ability to quickly identify key topics can empower global efforts to fight against an infodemic during a pandemic.

Correction: COVID-19 Discourse on Twitter in Four Asian Countries: Case Study of Risk Communication.

[This corrects the article DOI: 10.2196/23272.].

Integrating risk prediction models into chronic kidney disease care.

PURPOSE OF REVIEW: Although the concept of risk prediction in chronic kidney disease (CKD) is not new, how to integrate risk prediction models into CKD care remains largely unknown, particularly in the prevention and early management of CKD. The present review presents a timely overview of recent CKD risk prediction models and conceptualizes how these may be integrated into the care of patients with CKD. RECENT FINDINGS: In recent literature, prediction of time-to-ESKD has been thoroughly validated in multiple international cohorts, new models focused on CKD incidence, morbidity, and mortality have been developed, and ongoing work will determine the impact of integrating risk prediction models into CKD care on patients, nephrologists, and health systems. SUMMARY: With the availability of new models focused on CKD incidence, the United States Preventive Task Force should reconsider its determination of insufficient evidence for primary screening of CKD, which was due in part to the absence of validated risk models to guide CKD screening. Models predicting CKD morbidity and mortality present a new opportunity to standardize the intensity and frequency of care across nephrology practices.

Updating the International Index of Erectile Function: Evaluation of a Large Clinical Data Set.

INTRODUCTION: The International Index of Erectile Function (IIEF) is the predominant patient-reported outcomes instrument for assessing male sexual function. There are obvious problems with the use of the IIEF in the assessment of an individual patient, such as for men who use injections and men who do not engage in intercourse. AIM: The aim of the current study is to redesign the erectile function domain of the IIEF (IIEF6) to more accurately assess the individual patient. METHODS: In an observational study of men undergoing treatment for prostate cancer at a tertiary care institution, including 24,732 questionnaires completed by 6,780 individuals, IIEF6 scores were compared for patients using and not using erectile aids. Men not engaging in sexual intercourse were asked to describe the reason. MAIN OUTCOME MEASURE: The main outcome we were seeking was the IIEF6 scores. RESULTS: Mean scores before erectile aids items were added was 17.7 compared with predicted scores of 18.3 vs 16.7 if patients reported their function with vs without the use of aids. No intercourse was reported for 35% of surveys. Reasons given were lack of ability or confidence in 53%, lack of willing and available partner in 28%, "other" in 17% (including respondent's or partner's health issues, low libido, preference for nonpenetrative sex or for sex with men). Doubling the sum of the 3 nonintercourse IIEF6 questions had excellent properties (difference of 0.06, limits of agreement -3.10 to 3.22). CLINICAL IMPLICATIONS: Erectile function instruments must include items about erectile aids. Men who report that they have not attempted intercourse should not be assumed to have erectile dysfunction, but should be asked the reason why. For men who report lack of opportunity or preference for intercourse, the score of the 3 nonintercourse IIEF6 questions should be doubled. STRENGTHS & LIMITATIONS: This is a large study of patients in a real-world setting. Although the study only includes radical prostatectomy patients, and although the study cohort is not fully representative of the US prostate cancer population as a whole, these issues would not affect the key findings. CONCLUSION: The IIEF6 can be redesigned to better assess the individual patient. Vickers AJ, Tin AL, Singh K, et al. Updating the International Index of Erectile Function: Evaluation of a Large Clinical Data Set. J Sex Med 2020;17:126-132.

Prospective monitoring of imaging guideline adherence by physicians in a surgical collaborative: comparison of statistical process control methods for detecting outlying performance.

BACKGROUND: Systematic, automated methods for monitoring physician performance are necessary if outlying behavior is to be detected promptly and acted on. In the Michigan Urological Surgery Improvement Collaborative (MUSIC), we evaluated several statistical process control (SPC) methods to determine the sensitivity and ease of interpretation for assessing adherence to imaging guidelines for patients with newly diagnosed prostate cancer. METHODS: Following dissemination of imaging guidelines within the Michigan Urological Surgery Improvement Collaborative (MUSIC) for men with newly diagnosed prostate cancer, MUSIC set a target of imaging < 10% of patients for which bone scan is not indicated. We compared four SPC methods using Monte Carlo simulation: p-chart, weighted binomial CUSUM, Bernoulli cumulative sum (CUSUM), and exponentially weighted moving average (EWMA). We simulated non-indicated bone scan rates ranging from 5.9% (within target) to 11.4% (above target) for a representative MUSIC practice. Sensitivity was determined using the average run length (ARL), the time taken to signal a change. We then plotted actual non-indicated bone scan rates for a representative MUSIC practice using each SPC method to qualitatively assess graphical interpretation. RESULTS: EWMA had the lowest ARL and was able to detect changes significantly earlier than the other SPC methodologies (p < 0.001). The p-chart had the highest ARL and thus detected changes slowest (p < 0.001). EWMA and p-charts were easier to interpret graphically than CUSUM methods due to their ability to display historical imaging rates. CONCLUSIONS: SPC methods can be used to provide informative and timely feedback regarding adherence to healthcare performance target rates in quality improvement collaboratives. We found the EWMA method most suited for detecting changes in imaging utilization.

Iron, Hepcidin, and Death in Human AKI.

BACKGROUND: Iron is a key mediator of AKI in animal models, but data on circulating iron parameters in human AKI are limited. METHODS: We examined results from the ARF Trial Network study to assess the association of plasma catalytic iron, total iron, transferrin, ferritin, free hemoglobin, and hepcidin with 60-day mortality. Participants included critically ill patients with AKI requiring RRT who were enrolled in the study. RESULTS: Of the 807 study participants, 409 (51%) died by day 60. In both unadjusted and multivariable adjusted models, higher plasma concentrations of catalytic iron were associated with a significantly greater risk of death, as were lower concentrations of hepcidin. After adjusting for other factors, patients with catalytic iron levels in the highest quintile versus the lowest quintile had a 4.06-fold increased risk of death, and patients with hepcidin levels in the lowest quintile versus the highest quintile of hepcidin had a 3.87-fold increased risk of death. These findings were consistent across multiple subgroups. Other iron markers were also associated with death, but the magnitude of the association was greatest for catalytic iron and hepcidin. Higher plasma concentrations of catalytic iron and lower concentrations of hepcidin are each independently associated with mortality in critically ill patients with AKI requiring RRT. CONCLUSIONS: These findings suggest that plasma concentrations of catalytic iron and hepcidin may be useful prognostic markers in patients with AKI. Studies are needed to determine whether strategies to reduce catalytic iron or increase hepcidin might be beneficial in this patient population.

Grade Groups Provide Improved Predictions of Pathological and Early Oncologic Outcomes Compared with Gleason Score Risk Groups.

PURPOSE: The GG (Grade Group) system was introduced in 2013. Data from academic centers suggest that GG better distinguishes between prostate cancer risk groups than the Gleason score (GS) risk groups. We compared the performance of the 2 systems to predict pathological/recurrence outcomes using data from the MUSIC (Michigan Urological Surgery Improvement Collaborative). MATERIALS AND METHODS: Patients who underwent biopsy and radical prostatectomy in the MUSIC from March 2012 to June 2017 were classified according to GG and GS. Outcomes included the presence or absence of extraprostatic extension, seminal vesical invasion, positive lymph nodes, positive surgical margins and time to cancer recurrence (defined as postoperative prostate specific antigen 0.2 ng/ml or greater). Logistic and Cox regression models were used to compare the difference in outcomes. RESULTS: A total of 8,052 patients were identified. When controlling for patient characteristics, significantly higher risks of extraprostatic extension, seminal vesical invasion and positive lymph nodes were observed for biopsy GG 3 vs 2 and for GG 5 vs 4 (p <0.001). Biopsy GGs 3, 4 and 5 also showed shorter time to biochemical recurrence than GGs 2, 3 and 4, respectively (p <0.001). GGs 3, 4 and 5 at radical prostatectomy were each associated with a greater probability of recurrence compared to the next lower GG (p <0.001). GG (vs GS) had better predictive power for extraprostatic extension, seminal vesical invasion, positive lymph nodes and biochemical recurrence. CONCLUSIONS: GG at biopsy and radical prostatectomy allows for better discrimination of recurrence-free survival between individual risk groups than GS risk groups with GGs 2, 3, 4 and 5 each incrementally associated with increased risk.

Association of Chronic Kidney Disease with Preserved Ejection Fraction Heart Failure Is Independent of Baseline Cardiac Function.

BACKGROUND/AIMS: Chronic kidney disease (CKD) is common among patients with heart failure with preserved ejection fraction (HFpEF) and is associated with worse clinical outcomes. This study aims to identify whether the association of CKD with HFpEF is independent of underlying echocardiographic abnormalities. MATERIALS: We conducted a retrospective cohort study including patients without prevalent heart failure referred for echocardiography. Patients with serial echocardiograms, baseline left ventricular ejection fraction (LVEF) ≥50% and estimated glomerular filtration rate (eGFR) ≥90 mL/min/1.73 m2 were matched 1:1 with patients with eGFR <60 mL/min/1.73 m2 for age (±5 years), sex, history of hypertension or diabetes, use of renin-angiotensin inhibitors, and LVEF (±5%). A secondary analysis included patients with preserved LVEF and normal left ventricular mass index matched for the same parameters except use of renin-angiotensin inhibitors. RESULTS: Patients with CKD were at increased risk for HFpEF admission: crude hazard ratio (HR) 1.79 (95% confidence interval [CI] 1.38-2.32, p < 0.001) and adjusted HR (for coronary disease, loop diuretics, left atrial diameter) 1.64 (95% CI 1.22-2.21, p = 0.001). LVEF and left ventricular diameter decreased over time in both groups but no difference was observed in rate of dropping. Results were similar in the secondary analysis (crude HR 1.99, 95% CI 1.07-3.71, p = 0.03 and HR adjusted for left atrial diameter 1.98, 95% CI 1.05-3.75, p = 0.04). Rate of change was similar for LVEF, pulmonary artery pressure, and left ventricular mass index in both groups. CONCLUSION: CKD is independently associated with incident HFpEF despite a similar change in relevant echocardiographic parameters in patients with or without CKD.

A concept-wide association study to identify potential risk factors for nonadherence among prevalent users of antihypertensives.

PURPOSE: We sought to determine whether an association study using information contained in clinical notes could identify known and potentially novel risk factors for nonadherence to antihypertensive medications. METHODS: We conducted a retrospective concept-wide association study (CWAS) using clinical notes to identify potential risk factors for medication nonadherence, adjusting for age, sex, race, baseline blood pressure, estimated glomerular filtration rate, and a combined comorbidity score. Participants included Medicare beneficiaries 65 years and older receiving care at the Harvard Vanguard Medical Associates network from 2010-2012 and enrolled in a Medicare Advantage program. Concepts were extracted from clinical notes in the year prior to the index prescription date for each patient. We tested associations with the outcome for 5013 concepts extracted from clinical notes in a derivation cohort (4382 patients) and accounted for multiple hypothesis testing by using a false discovery rate threshold of less than 5% (q < .05). We then confirmed the associations in a validation cohort (3836 patients). Medication nonadherence was defined using a proportion of days covered (PDC) threshold less than 0.8 using pharmacy claims data. RESULTS: We found 415 concepts associated with nonadherence, which we organized into 11 clusters using a hierarchical clustering approach. Volume depletion and overload, assessment of needs at the point of discharge, mood disorders, neurological disorders, complex coordination of care, and documentation of noncompliance were some of the factors associated with nonadherence. CONCLUSIONS: This approach was successful in identifying previously described and potentially new risk factors for antihypertensive nonadherence using the clinical narrative.

Complexities, variations, and errors of numbering within clinical notes: the potential impact on information extraction and cohort-identification.

BACKGROUND: Numbers and numerical concepts appear frequently in free text clinical notes from electronic health records. Knowledge of the frequent lexical variations of these numerical concepts, and their accurate identification, is important for many information extraction tasks. This paper describes an analysis of the variation in how numbers and numerical concepts are represented in clinical notes. METHODS: We used an inverted index of approximately 100 million notes to obtain the frequency of various permutations of numbers and numerical concepts, including the use of Roman numerals, numbers spelled as English words, and invalid dates, among others. Overall, twelve types of lexical variants were analyzed. RESULTS: We found substantial variation in how these concepts were represented in the notes, including multiple data quality issues. We also demonstrate that not considering these variations could have substantial real-world implications for cohort identification tasks, with one case missing > 80% of potential patients. CONCLUSIONS: Numbering within clinical notes can be variable, and not taking these variations into account could result in missing or inaccurate information for natural language processing and information retrieval tasks.

The relative benefits of claims and electronic health record data for predicting medication adherence trajectory.

BACKGROUND: Healthcare providers are increasingly encouraged to improve their patients' adherence to chronic disease medications. Prediction of adherence can identify patients in need of intervention, but most prediction efforts have focused on claims data, which may be unavailable to providers. Electronic health records (EHR) are readily available and may provide richer information with which to predict adherence than is currently available through claims. METHODS: In a linked database of complete Medicare Advantage claims and comprehensive EHR from a multi-specialty outpatient practice, we identified patients who filled a prescription for a statin, antihypertensive, or oral antidiabetic during 2011 to 2012. We followed patients to identify subsequent medication filling patterns and used group-based trajectory models to assign patients to adherence trajectories. We then identified potential predictors from both claims and EHR data and fit a series of models to evaluate the accuracy of each data source in predicting medication adherence. RESULTS: Claims were highly predictive of patients in the worst adherence trajectory (C=0.78), but EHR data also provided good predictions (C=0.72). Among claims predictors, presence of a prior gap in filling of at least 6 days was by far the most influential predictor. In contrast, good predictions from EHR data required complex models with many variables. CONCLUSION: EHR data can provide good predictions of adherence trajectory and therefore may be useful for providers seeking to deploy resource-intensive interventions. However, prior adherence information derived from claims is most predictive, and can supplement EHR data when it is available.

Catalyst-Controlled Structural Divergence: Selective Intramolecular 7-endo-dig and 6-exo-dig Post-Ugi Cyclization for the Synthesis of Benzoxazepinones and Benzoxazinones.

Metal catalyzed post-Ugi cyclization of bis-amides is reported in this study. Exposure of bis-amides to Pd(II) catalyst triggered the formation of seven-membered benzoxazepinones. This investigation established that changing the catalyst to a Echavarren's gold(I) turned off cyclization to seven member ring and turned on 6-exo-dig annulations to afford family of six-membered benzoxazinones. To support the proposed mechanisms, quantum chemical based density functional theory calculations have been performed and validated. This novel method obtained molecular complexity up to four modular inputs and divergence of two different skeletons. 2D NMR spectroscopic techniques and single crystal X-ray diffraction established the proposed structures.