Identification of a potential fibromyalgia diagnosis using random forest modeling applied to electronic medical records

Uncovering the burden of hidradenitis suppurativa misdiagnosis and underdiagnosis: a machine learning approach

Hidradenitis suppurativa (HS) is a chronic inflammatory follicular skin condition that is associated with significant psychosocial and economic burden and a diminished quality of life and work productivity. Accurate diagnosis of HS is challenging due to its unknown etiology, which can lead to underdiagnosis or misdiagnosis that results in increased patient and healthcare system burden. We applied machine learning (ML) to a medical and pharmacy claims database using data from 2000 through 2018 to develop a novel model to better understand HS underdiagnosis on a healthcare system level. The primary results demonstrated that high-performing models for predicting HS diagnosis can be constructed using claims data, with an area under the curve (AUC) of 81%-82% observed among the top-performing models. The results of the models developed in this study could be input into the development of an impact of inaction model that determines the cost implications of HS diagnosis and treatment delay to the healthcare system.

A machine learning model for diagnosing acute pulmonary embolism and comparison with Wells score, revised Geneva score, and Years algorithm

BACKGROUND

Acute pulmonary embolism (APE) is a fatal cardiovascular disease, yet missed diagnosis and misdiagnosis often occur due to non-specific symptoms and signs. A simple, objective technique will help clinicians make a quick and precise diagnosis. In population studies, machine learning (ML) plays a critical role in characterizing cardiovascular risks, predicting outcomes, and identifying biomarkers. This work sought to develop an ML model for helping APE diagnosis and compare it against current clinical probability assessment models.

METHODS

This is a single-center retrospective study. Patients with suspected APE were continuously enrolled and randomly divided into two groups including training and testing sets. A total of 8 ML models, including random forest (RF), Naïve Bayes, decision tree, K-nearest neighbors, logistic regression, multi-layer perceptron, support vector machine, and gradient boosting decision tree were developed based on the training set to diagnose APE. Thereafter, the model with the best diagnostic performance was selected and evaluated against the current clinical assessment strategies, including the Wells score, revised Geneva score, and Years algorithm. Eventually, the ML model was internally validated to assess the diagnostic performance using receiver operating characteristic (ROC) analysis.

RESULTS

The ML models were constructed using eight clinical features, including D-dimer, cardiac troponin T (cTNT), arterial oxygen saturation, heart rate, chest pain, lower limb pain, hemoptysis, and chronic heart failure. Among eight ML models, the RF model achieved the best performance with the highest area under the curve (AUC) (AUC = 0.774). Compared to the current clinical assessment strategies, the RF model outperformed the Wells score ( P = 0.030) and was not inferior to any other clinical probability assessment strategy. The AUC of the RF model for diagnosing APE onset in internal validation set was 0.726.

CONCLUSIONS

Based on RF algorithm, a novel prediction model was finally constructed for APE diagnosis. When compared to the current clinical assessment strategies, the RF model achieved better diagnostic efficacy and accuracy. Therefore, the ML algorithm can be a useful tool in assisting with the diagnosis of APE.

Identifying persistent somatic symptoms in electronic health records: exploring multiple theory-driven methods of identification

OBJECTIVE

Persistent somatic symptoms (PSSs) are defined as symptoms not fully explained by well-established pathophysiological mechanisms and are prevalent in up to 10% of patients in primary care. The present study aimed to explore methods to identify patients with a recognisable risk of having PSS in routine primary care data.

DESIGN

A cross-sectional study to explore four identification methods that each cover part of the broad spectrum of PSS was performed. Cases were selected based on (1) PSS-related syndrome codes, (2) PSS-related symptom codes, (3) PSS-related terminology and (4) Four-Dimensional Symptom Questionnaire scores and all methods combined.

SETTING

Coded electronic health record data were extracted from 76 general practices in the Netherlands.

PARTICIPANTS

Patients who were registered for at least 1 year during 2014-2018, were included (n=169 138).

OUTCOME MEASURES

Identification methods were explored based on (1) PSS sample sizes and demographics, (2) presence of chronic conditions and (3) healthcare utilisation (HCU) variables. Overlap between methods and practice specific differences were examined.

RESULTS

The percentage of cases identified varied between 0.3% and 7.0% across the methods. Over 58.1% of cases had chronic physical condition(s) and over 33.8% had chronic mental condition(s). HCU was generally higher for cases selected by any method compared with the total cohort. HCU was higher for method B compared with the other methods. In 26.7% of cases, cases were selected by multiple methods. Overlap between methods was low.

CONCLUSIONS

Different methods yielded different patient samples which were general practice specific. Therefore, for the most comprehensive data-based selection of PSS cases, a combination of methods A, C and D would be recommended. Advanced (data-driven) methods are needed to create a more sensitive algorithm for identifying the full spectrum of PSS. For clinical purposes, method B could possibly support screening of patients who are currently missed in daily practice.

Racial Differences in the Use of Aortic Valve Replacement for Treatment of Symptomatic Severe Aortic Valve Stenosis in the Transcatheter Aortic Valve Replacement Era

Background Aortic valve replacement (AVR) is a life-saving treatment for patients with symptomatic severe aortic valve stenosis. We sought to determine whether transcatheter AVR has resulted in a more equitable treatment rate by race in the United States. Methods and Results A total of 32 853 patients with symptomatic severe aortic valve stenosis were retrospectively identified via Optum's deidentified electronic health records database (2007-2017). AVR rates in non-Hispanic Black and White patients were assessed in the year after diagnosis. Multivariate Fine-Gray hazards models were used to evaluate the likelihood of AVR by race, with adjustment for patient factors and the managing cardiologist. Time-trend and 1-year symptomatic severe aortic valve stenosis survival analyses were also performed. From 2011 to 2016, the rate of AVR increased from 20.1% to 37.1%. Overall, Black individuals were less likely than Whites to receive AVR (22.9% versus 31.0%; unadjusted hazard ratio [HR], 0.70; 95% CI, 0.62-0.79; fully adjusted HR, 0.76; 95% CI, 0.67-0.85). Yet, during 2015 to 2016, AVR racial differences were attenuated (29.5% versus 35.2%; adjusted HR, 0.86; 95% CI, 0.74-1.02) because of greater uptake of transcatheter AVR in Blacks than Whites (53.4% of AVRs versus 47.3%; P=0.128). Untreated patients had significantly higher 1-year mortality than those treated (adjusted HR, 0.57; 95% CI, 0.53-0.61), which was consistent by race (interaction P value=0.52). Conclusions Although transcatheter AVR has increased the use of AVR in the United States, treatment rates remain low. Black patients with symptomatic severe aortic valve stenosis were less likely than White patients to receive AVR, yet these differences have recently narrowed.

Opioid Prescribing and Polypharmacy in Children with Chronic Musculoskeletal Pain

OBJECTIVE

Thirty percent of adults with fibromyalgia receive an opioid, but the prevalence of opioid prescribing in pediatric chronic musculoskeletal pain is unknown. The aims of this study were to determine the prevalence of and factors associated with opioid exposure and polypharmacy among children with chronic musculoskeletal pain.

METHODS

In this retrospective cohort study using health care claims data from 2000 to 2013, the index date was the first ICD-9 code 729.1. Included subjects were ≥ 2 and < 18 years old at the index date with two or more codes within 12 months and 18 months of continuous enrollment. Subjects with burns, sickle cell disease, or malignancy were excluded. Opioid exposure was defined as one or more prescriptions within six months before or any time after the index date. Polypharmacy was considered minor (2-4 medications) or major (≥5 medications).

RESULTS

Of 25,321 included subjects, 20% received an opioid and 26% experienced minor polypharmacy. Opioid exposure was associated with female sex (odds ratio [OR] = 1.27, P < 0.01), Caucasian race (OR = 1.27, P < 0.01), hospitalization (OR = 1.20, P < 0.01), and visit with anesthesiology (OR = 1.97, P < 0.01) or orthopedics (OR = 1.09, P < 0.05). Mental health codes were associated with decreased odds of opioid exposure (all P < 0.05). Children seen by a chiropractor or physiatrist had a reduced odds of receipt of an opioid (OR = 0.42 and 0.84, respectively, both P < 0.01).

CONCLUSIONS

Twenty percent of children with chronic musculoskeletal pain received an opioid. Twenty-six percent experienced polypharmacy, with the majority receiving 2-4 medications. Increased availability of psychological and nonpharmacologic services are potential strategies to reduce opioid exposure.

Characterizing classes of fibromyalgia within the continuum of central sensitization syndrome

Background

While fibromyalgia (FM) is characterized by chronic widespread pain and tenderness, its presentation among patients as a continuum of diseases rather than a single disease contributes to the challenges of diagnosis and treatment. The purpose of this analysis was to distinguish and characterize classes of FM within the continuum using data from chronic pain patients.

Methods

FM patients were identified from administrative claims data from the ProCare Systems’ network of Michigan pain clinics between January 1999 and February 2015. Identification was based on either use of traditional criteria (ie, ICD-9 codes) or a predictive model indicative of patients having FM. Patients were classified based on similarity of comorbidities (symptom severity), region of pain (widespread pain), and type and number of procedures (treatment intensity) using unsupervised learning. Text mining and a review of physician notes were conducted to assist in understanding the FM continuum.

Results

A total of 2,529 FM patients with 79,570 observations or clinical visits were evaluated. Four main classes of FM patients were identified: Class 1) regional FM with classic symptoms; Class 2) generalized FM with increasing widespread pain and some additional symptoms; Class 3) FM with advanced and associated conditions, increasing widespread pain, increased sleep disturbance, and chemical sensitivity; and Class 4) FM secondary to other conditions.

Conclusion

FM is a disease continuum characterized by progressive and identifiable classifications. Four classes of FM can be differentiated by pain and symptom severity, specific comorbidities, and use of clinical procedures.

Evaluating increased resource use in fibromyalgia using electronic health records

Objective

The management of fibromyalgia (FM), a chronic musculoskeletal disease, remains challenging, and patients with FM are often characterized by high health care resource utilization. This study sought to explore potential drivers of all-cause health care resource utilization and other factors associated with high resource use, using a large electronic health records (EHR) database to explore data from patients diagnosed with FM.

Methods

This was a retrospective analysis of de-identified EHR data from the Humedica database. Adults (≥18 years) with FM were identified based on ≥2 International Classification of Diseases, Ninth Revision codes for FM (729.1) ≥30 days apart between January 1, 2008 and December 31, 2012 and were required to have evidence of ≥12 months continuous care pre- and post-index; first FM diagnosis was the index event; 12-month pre- and post-index reporting periods. Multivariable analysis evaluated relationships between variables and resource utilization.

Results

Patients were predominantly female (81.4%), Caucasian (87.7%), with a mean (standard deviation) age of 54.4 (14.8) years. The highest health care resource utilization was observed for the categories of “medication orders” and “physician office visits,” with 12-month post-index means of 21.2 (21.5) drug orders/patient and 15.1 (18.1) office visits/patient; the latter accounted for 73.3% of all health care visits. Opioids were the most common prescription medication, 44.3% of all patients. The chance of high resource use was significantly increased (P<0.001) 26% among African-Americans vs Caucasians and for patients with specific comorbid conditions ranging from 6% (musculoskeletal pain or depression/bipolar disorder) to 21% (congestive heart failure). Factors significantly associated with increased medications ordered included being female (P<0.001) and specific comorbid conditions (P<0.05).

Conclusion

Physician office visits and pharmacotherapy orders were key drivers of all-cause health care utilization, with demographic factors, opioid use, and specific comorbidities associated with resource intensity. Health systems and providers may find their EHRs to be a useful tool for identifying and managing resource-intensive FM patients.