Text Mining of Electronic Health Records Can Accurately Identify and Characterize Patients With Systemic Lupus Erythematosus

Association between systemic lupus erythematosus and myasthenia gravis: A population-based National Study

BACKGROUND

Systemic lupus erythematosus (SLE) and myasthenia gravis (MG) are autoimmune diseases. Previous case reports and case series suggest an association may exist between these diseases, as well as an increased risk of SLE after thymectomy for MG. We undertook this study to determine whether SLE and MG were associated in large cohorts.

METHODS

We searched the IBM Watson Health Explorys platform and the Department of Veterans Affairs Million Veteran Program (MVP) database for diagnoses of SLE and MG. In addition, we examined subjects enrolled in the Lupus Family Registry and Repository (LFRR) as well as controls for a diagnosis of MG.

RESULTS

Among 59,780,210 individuals captured in Explorys, there were 25,750 with MG and 65,370 with SLE. 370 subjects had both. Those with MG were >10 times more likely to have SLE than those without MG. Those with both diseases were more likely to be women, African American, and at a younger age than MG subjects without SLE. In addition, the MG patients who underwent thymectomy had an increased risk of SLE compared to MG patients who had not undergone thymectomy (OR 3.11, 95% CI: 2.12 to 4.55). Autoimmune diseases such as pernicious anemia and miscellaneous comorbidities such as chronic kidney disease were significantly more common in MG patients who developed SLE. In the MVP, SLE and MG were also significantly associated. Association of SLE and MG in a large SLE cohort with rigorous SLE classification confirmed the association of SLE with MG at a similar level.

CONCLUSION

While the number of patients with both MG and SLE is small, SLE and MG are strongly associated together in very large databases and a large SLE cohort.

Using an artificial intelligence tool incorporating natural language processing to identify patients with a diagnosis of ANCA-associated vasculitis in electronic health records

BACKGROUND

Because anti-neutrophil cytoplasmatic antibody (ANCA)-associated vasculitis (AAV) is a rare, life-threatening, auto-immune disease, conducting research is difficult but essential. A long-lasting challenge is to identify rare AAV patients within the electronic-health-record (EHR)-system to facilitate real-world research. Artificial intelligence (AI)-search tools using natural language processing (NLP) for text-mining are increasingly postulated as a solution.

METHODS

We employed an AI-tool that combined text-mining with NLP-based exclusion, to accurately identify rare AAV patients within large EHR-systems (>2.000.000 records). We developed an identification method in an academic center with an established AAV-training set (n = 203) and validated the method in a non-academic center with an AAV-validation set (n = 84). To assess accuracy anonymized patient records were manually reviewed.

RESULTS

Based on an iterative process, a text-mining search was developed on disease description, laboratory measurements, medication and specialisms. In the training center, 608 patients were identified with a sensitivity of 97.0 % (95%CI [93.7, 98.9]) and positive predictive value (PPV) of 56.9 % (95%CI [52.9, 60.1]). NLP-based exclusion resulted in 444 patients increasing PPV to 77.9 % (95%CI [73.7, 81.7]) while sensitivity remained 96.3 % (95%CI [93.8, 98.0]). In the validation center, text-mining identified 333 patients (sensitivity 97.6 % (95%CI [91.6, 99.7]), PPV 58.2 % (95%CI [52.8, 63.6])) and NLP-based exclusion resulted in 223 patients, increasing PPV to 86.1 % (95%CI [80.9, 90.4]) with 98.0 % (95%CI [94.9, 99.4]) sensitivity. Our identification method outperformed ICD-10-coding predominantly in identifying MPO+ and organ-limited AAV patients.

CONCLUSIONS

Our study highlights the advantages of implementing AI, notably NLP, to accurately identify rare AAV patients within large EHR-systems and demonstrates the applicability and transportability. Therefore, this method can reduce efforts to identify AAV patients and accelerate real-world research, while avoiding bias by ICD-10-coding.

Scalable and interpretable alternative to chart review for phenotype evaluation using standardized structured data from electronic health records

OBJECTIVES

Chart review as the current gold standard for phenotype evaluation cannot support observational research on electronic health records and claims data sources at scale. We aimed to evaluate the ability of structured data to support efficient and interpretable phenotype evaluation as an alternative to chart review.

MATERIALS AND METHODS

We developed Knowledge-Enhanced Electronic Profile Review (KEEPER) as a phenotype evaluation tool that extracts patient's structured data elements relevant to a phenotype and presents them in a standardized fashion following clinical reasoning principles. We evaluated its performance (interrater agreement, intermethod agreement, accuracy, and review time) compared to manual chart review for 4 conditions using randomized 2-period, 2-sequence crossover design.

RESULTS

Case ascertainment with KEEPER was twice as fast compared to manual chart review. 88.1% of the patients were classified concordantly using charts and KEEPER, but agreement varied depending on the condition. Missing data and differences in interpretation accounted for most of the discrepancies. Pairs of clinicians agreed in case ascertainment in 91.2% of the cases when using KEEPER compared to 76.3% when using charts. Patient classification aligned with the gold standard in 88.1% and 86.9% of the cases respectively.

CONCLUSION

Structured data can be used for efficient and interpretable phenotype evaluation if they are limited to relevant subset and organized according to the clinical reasoning principles. A system that implements these principles can achieve noninferior performance compared to chart review at a fraction of time.

Use of electronic health record data mining for heart failure subtyping

OBJECTIVE

To assess whether electronic health record (EHR) data text mining can be used to improve register-based heart failure (HF) subtyping. EHR data of 43,405 individuals from two Finnish hospital biobanks were mined for unstructured text mentions of ejection fraction (EF) and validated against clinical assessment in two sets of 100 randomly selected individuals. Structured laboratory data was then incorporated for a categorization by HF subtype (HF with mildly reduced EF, HFmrEF; HF with preserved EF, HFpEF; HF with reduced EF, HFrEF; and no HF).

RESULTS

In 86% of the cases, the algorithm-identified EF belonged to the correct HF subtype range. Sensitivity, specificity, PPV and NPV of the algorithm were 94-100% for HFrEF, 85-100% for HFmrEF, and 96%, 67%, 53% and 98% for HFpEF. Survival analyses using the traditional diagnosis of HF were in concordance with the algorithm-based ones. Compared to healthy individuals, mortality increased from HFmrEF (hazard ratio [HR], 1.91; 95% confidence interval [CI], 1.24-2.95) to HFpEF (2.28; 1.80-2.88) to HFrEF group (2.63; 1.97-3.50) over a follow-up of 1.5 years. We conclude that quantitative EF data can be efficiently extracted from EHRs and used with laboratory data to subtype HF with reasonable accuracy, especially for HFrEF.

Early Detection of SARS-CoV-2 Epidemic Waves: Lessons from the Syndromic Surveillance in Lombardy, Italy

We evaluated the performance of the exponentially weighted moving average (EWMA) model for comparing two families of predictors (i.e., structured and unstructured data from visits to the emergency department (ED)) for the early detection of SARS-CoV-2 epidemic waves. The study included data from 1,282,100 ED visits between 1 January 2011 and 9 December 2021 to a local health unit in Lombardy, Italy. A regression model with an autoregressive integrated moving average (ARIMA) error term was fitted. EWMA residual charts were then plotted to detect outliers in the frequency of the daily ED visits made due to the presence of a respiratory syndrome (based on coded diagnoses) or respiratory symptoms (based on free text data). Alarm signals were compared with the number of confirmed SARS-CoV-2 infections. Overall, 150,300 ED visits were encoded as relating to respiratory syndromes and 87,696 to respiratory symptoms. Four strong alarm signals were detected in March and November 2020 and 2021, coinciding with the onset of the pandemic waves. Alarm signals generated for the respiratory symptoms preceded the occurrence of the first and last pandemic waves. We concluded that the EWMA model is a promising tool for predicting pandemic wave onset.

Data mining to retrieve smoking status from electronic health records in general practice. EUROPEAN HEART JOURNAL. DIGITAL HEALTH 2022;3:437-444. [PMID: 36712169 PMCID: PMC9707867 DOI: 10.1093/ehjdh/ztac031]

Aims

Optimize and assess the performance of an existing data mining algorithm for smoking status from hospital electronic health records (EHRs) in general practice EHRs.

Methods and results

We optimized an existing algorithm in a training set containing all clinical notes from 498 individuals (75 712 contact moments) from the Julius General Practitioners' Network (JGPN). Each moment was classified as either 'current smoker', 'former smoker', 'never smoker', or 'no information'. As a reference, we manually reviewed EHRs. Algorithm performance was assessed in an independent test set (n = 494, 78 129 moments) using precision, recall, and F1-score. Test set algorithm performance for 'current smoker' was precision 79.7%, recall 78.3%, and F1-score 0.79. For former smoker, it was precision 73.8%, recall 64.0%, and F1-score 0.69. For never smoker, it was precision 92.0%, recall 74.9%, and F1-score 0.83. On a patient level, performance for ever smoker (current and former smoker combined) was precision 87.9%, recall 94.7%, and F1-score 0.91. For never smoker, it was 98.0, 82.0, and 0.89%, respectively. We found a more narrative writing style in general practice than in hospital EHRs.

Conclusion

Data mining can successfully retrieve smoking status information from general practice clinical notes with a good performance for classifying ever and never smokers. Differences between general practice and hospital EHRs call for optimization of data mining algorithms when applied beyond a primary development setting.

Microarray testing in patients with systemic lupus erythematosus identifies a high prevalence of CpG DNA-binding antibodies

Objective

Many autoantibodies are known to be associated with SLE, although their role in clinical practice is limited because of low sensitivity and weak associations with clinical manifestations. There has been great interest in the discovery of new autoantibodies to use in clinical practice. In this study, we investigated 57 new and known antibodies and their potential for diagnostics or risk stratification.

Methods

Between 2014 and 2017, residual sera of all anti-dsDNA tests in the UMC Utrecht were stored in a biobank. This included sera of patients with SLE, patients with a diagnosis of another immune-mediated inflammatory disease (IMID), patients with low (non-IMID) or medium levels of clinical suspicion of SLE but no IMID diagnosis (Rest), and self-reported healthy blood bank donors. Diagnosis and (presence of) symptoms at each blood draw were retrospectively assessed in the patient records with the Utrecht Patient-Oriented Database using a newly developed text mining algorithm. Sera of patients were analysed for the presence of 57 autoantibodies with a custom-made immunofluorescent microarray. Signal intensity cut-offs for all antigens on the microarray were set to the 95th percentile of the non-IMID control group. Differences in prevalence of autoantibodies between patients with SLE and control groups were assessed.

Results

Autoantibody profiles of 483 patients with SLE were compared with autoantibody profiles of 1397 patients from 4 different control groups. Anti-dsDNA was the most distinguishing feature between patients with SLE and other patients, followed by antibodies against Cytosine-phosphate-Guanine (anti-CpG) DNA motifs (p<0.0001). Antibodies against CMV (cytomegalovirus) and ASCA (anti-Saccharomyces cerevisiae antibodies) were more prevalent in patients with SLE with (a history of) lupus nephritis than patients with SLE without nephritis.

Conclusion

Antibodies against CpG DNA motifs are prevalent in patients with SLE. Anti-CMV antibodies are associated with lupus nephritis.

Digital health, big data and smart technologies for the care of patients with systemic autoimmune diseases: Where do we stand?

The past decade has seen tremendous development in digital health, including in innovative new technologies such as Electronic Health Records, telemedicine, virtual visits, wearable technology and sophisticated analytical tools such as artificial intelligence (AI) and machine learning for the deep-integration of big data. In the field of rare connective tissue diseases (rCTDs), these opportunities include increased access to scarce and remote expertise, improved patient monitoring, increased participation and therapeutic adherence, better patient outcomes and patient empowerment. In this review, we discuss opportunities and key-barriers to improve application of digital health technologies in the field of autoimmune diseases. We also describe what could be the fully digital pathway of rCTD patients. Smart technologies can be used to provide real-world evidence about the natural history of rCTDs, to determine real-life drug utilization, advanced efficacy and safety data for rare diseases and highlight significant unmet needs. Yet, digitalization remains one of the most challenging issues faced by rCTD patients, their physicians and healthcare systems. Digital health technologies offer enormous potential to improve autoimmune rCTD care but this potential has so far been largely unrealized due to those significant obstacles. The need for robust assessments of the efficacy, affordability and scalability of AI in the context of digital health is crucial to improve the care of patients with rare autoimmune diseases.