Text Mining in Electronic Medical Records Enables Quick and Efficient Identification of Pregnancy Cases Occurring After Breast Cancer

Bibliometric Analysis of Development Trends and Research Hotspots in the Study of Data Mining in Nursing Based on CiteSpace

Backgrounds

With the advent of the big data era, hospital information systems and mobile care systems, among others, generate massive amounts of medical data. Data mining, as a powerful information processing technology, can discover non-obvious information by processing large-scale data and analyzing them in multiple dimensions. How to find the effective information hidden in the database and apply it to nursing clinical practice has received more and more attention from nursing researchers.

Aim

To look over the articles on data mining in nursing, compiled research status, identified hotspots, highlighted research trends, and offer recommendations for how data mining technology might be used in the nursing area going forward.

Methods

Data mining in nursing publications published between 2002 and 2023 were taken from the Web of Science Core Collection. CiteSpace was utilized for reviewing the number of articles, countries/regions, institutions, journals, authors, and keywords.

Results

According to the findings, the pace of data mining in nursing progress is not encouraging. Nursing data mining research is dominated by the United States and China. However, no consistent core group of writers or organizations has emerged in the field of nursing data mining. Studies on data mining in nursing have been increasingly gradually conducted in the 21st century, but the overall number is not large. Institution of Columbia University, journal of Cin-computers Informatics Nursing, author Diana J Wilkie, Muhammad Kamran Lodhi, Yingwei Yao are most influential in nursing data mining research. Nursing data mining researchers are currently focusing on electronic health records, text mining, machine learning, and natural language processing. Future research themes in data mining in nursing most include nursing informatics and clinical care quality enhancement.

Conclusion

Research data shows that data mining gives more perspectives for the growth of the nursing discipline and encourages the discipline's development, but it also introduces a slew of new issues that need researchers to address.

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.

Dysthyroidism during immune checkpoint inhibitors is associated with improved overall survival in adult cancers: data mining of 1385 electronic patient records

BACKGROUND

Dysthyroidism (DT) is a common toxicity of immune checkpoint inhibitors (ICIs) and prior work suggests that dysthyroidism (DT) might be associated with ICI efficacy.

PATIENTS AND METHODS

ConSoRe, a new generation data mining solution, was used in this retrospective study, to extract data from electronic patient records of adult cancer patients treated with ICI at Institut Paoli-Calmettes (Marseille, France). Every DT was verified and only ICI-induced DT was retained. Survival analyses were performed by Kaplan-Meier method (log-rank test) and Cox model. To account for immortal time bias, a conditional landmark analysis was performed (2 months and 6 months), together with a time-varying Cox model.

RESULTS

Data extraction identified 1385 patients treated with ICI between 2011 and 2021. DT was associated with improved overall survival (OS) (HR 0.46, (95% CI 0.33 to 0.65), p<0.001), with a median OS of 35.3 months in DT group vs 15.4 months in non-DT group (NDT). Survival impact of DT was consistent using a 6-month landmark analysis with a median OS of 36.7 months (95% CI 29.4 to not reported) in the DT group vs 25.5 months (95% CI 22.8 to 27.8) in the NDT group. In multivariate analysis, DT was independently associated with improved OS (HR 0.49, 95% CI 0.35 to 0.69, p=0.001). After adjustment in time-varying Cox model, this association remained significant (adjusted HR 0.64, 95% CI 0.45 to 0.90, p=0.010). Moreover, patients with DT and additional immune-related adverse event had increased OS compared with patients with isolated DT, with median OS of 38.8 months vs 21.4 months, respectively.

CONCLUSION

Data mining identified a large number of patients with ICI-induced DT, which was associated with improved OS accounting for immortal time bias.

An Application of Machine Learning Techniques to Analyze Patient Information to Improve Oral Health Outcomes

Objective

Various health-related fields have applied Machine learning (ML) techniques such as text mining, topic modeling (TM), and artificial neural networks (ANN) to automate tasks otherwise completed by humans to enhance patient care. However, research in dentistry on the integration of these techniques into the clinic arena has yet to exist. Thus, the purpose of this study was to: introduce a method of automating the reviewing patient chart information using ML, provide a step-by-step description of how it was conducted, and demonstrate this method's potential to identify predictive relationships between patient chart information and important oral health-related contributors.

Methods

A secondary data analysis was conducted to demonstrate the approach on a set of anonymized patient charts collected from a dental clinic. Two ML applications for patient chart review were demonstrated: (1) text mining and Latent Dirichlet Allocation (LDA) were used to preprocess, model, and cluster data in a narrative format and extract common topics for further analysis, (2) Ordinal logistic regression (OLR) and ANN were used to determine predictive relationships between the extracted patient chart data topics and oral health-related contributors. All analysis was conducted in R and SPSS (IBM, SPSS, statistics 22).

Results

Data from 785 patient charts were analyzed. Preprocessing of raw data (data cleaning and categorizing) identified 66 variables, of which 45 were included for analysis. Using LDA, 10 radiographic findings topics and 8 treatment planning topics were extracted from the data. OLR showed that caries risk, occlusal risk, biomechanical risk, gingival recession, periodontitis, gingivitis, assisted mouth opening, and muscle tenderness were highly predictable using the extracted radiographic and treatment planning topics and chart information. Using the statistically significant predictors obtained from OLR, ANN analysis showed that the model can correctly predict &gt;72% of all variables except for bruxism and tooth crowding (63.1 and 68.9%, respectively).

Conclusion

Our study presents a novel approach to address the need for data-enabled innovations in the field of dentistry and creates new areas of research in dental analytics. Utilizing ML methods and its application in dental practice has the potential to improve clinicians' and patients' understanding of the major factors that contribute to oral health diseases/conditions.

Agreement between blood pressure from research study visits versus electronic medical records and associations with hypertensive disorder diagnoses in pregnant women with overweight/obesity

OBJECTIVES

Blood pressure (BP) abstracted from electronic medical records (EMR) is moderately correlated to BP in nonpregnant adults with limited agreement. Little is known about the agreement of research versus EMR BP measured during pregnancy or associations of EMR BP with hypertensive disorder of pregnancy (HDP) diagnoses.

METHODS

BP was measured according to guidelines at in-person research study visits in 214 women with prepregnancy overweight or obesity (44.4% African American, mean age = 29.8 ± 4.8 years) at weeks 16 and 32 of pregnancy. Clinic BP readings that occurred within 1 week of the study visits were abstracted from the EMR. We assessed correlations between sources using Pearson's coefficients; the agreement was evaluated with Bland-Altman plots. We compared differences in the proportion of women with an HDP diagnosis in the EMR between women with versus without a hypertensive EMR BP measurement.

RESULTS

SBP and DBP from study visits and the EMR were modestly moderately correlated at both time points; 0.20 < r < 0.44; P < 0.05 for all. The average mean difference was 10.5 mmHg for SBP and <1 mmHg for DBP in early and 7.3 mmHg for SBP and -1.7 mmHg for DBP in late pregnancy. Women with at least one hypertensive BP reading in the EMR were more likely to have an HDP diagnosis recorded in the EMR; 43.5 versus 3.3%; P < 0.01.

CONCLUSION

EMR SBP was higher but moderately correlated with research quality BP in early and late pregnancy. Women with a hypertensive EMR BP measurement were more likely to have an HDP diagnosis in the EMR.

Factors Associated With the Discussion of Fertility Preservation in a Cohort of 1,357 Young Breast Cancer Patients Receiving Chemotherapy

PURPOSE

Female breast cancer (BC) patients exposed to gonadotoxic chemotherapy are at risk of future infertility. There is evidence of disparities in the discussion of fertility preservation for these patients. The aim of the study was to identify factors influencing the discussion of fertility preservation (FP).

MATERIAL AND METHODS

We analyzed consecutive BC patients treated by chemotherapy at Institut Curie from 2011-2017 and aged 18-43 years at BC diagnosis. The discussion of FP was classified in a binary manner (discussion/no discussion), based on mentions present in the patient's electronic health record (EHR) before the initiation of chemotherapy. The associations between FP discussion and the characteristics of patients/tumors and healthcare practitioners were investigated by logistic regression analysis.

RESULTS

The median age of the 1357 patients included in the cohort was 38.7 years, and median tumor size was 30.3 mm. The distribution of BC subtypes was as follows: 702 luminal BCs (58%), 241 triple-negative breast cancers (TNBCs) (20%), 193 HER2+/HR+ (16%) and 81 HER2+/HR- (6%). All patients received chemotherapy in a neoadjuvant (n=611, 45%) or adjuvant (n= 744, 55%) setting. A discussion of FP was mentioned for 447 patients (33%). Earlier age at diagnosis (discussion: 34.4 years versus no discussion: 40.5 years), nulliparity (discussion: 62% versus no discussion: 38%), and year of BC diagnosis were the patient characteristics significantly associated with the mention of FP discussion. Surgeons and female physicians were the most likely to mention FP during the consultation before the initiation of chemotherapy (discussion: 22% and 21%, respectively). The likelihood of FP discussion increased significantly over time, from 15% in 2011 to 45% in 2017. After multivariate analysis, FP discussion was significantly associated with younger age, number of children before BC diagnosis, physicians' gender and physicians' specialty.

CONCLUSION

FP discussion rates are low and are influenced by patient and physician characteristics. There is therefore room for improvement in the promotion and systematization of FP discussion.

Research and Application of Artificial Intelligence (AI) based on Electronic Health Records from Patients with Cancer: a Systematic Review (Preprint)

Background

With the accumulation of electronic health records and the development of artificial intelligence, patients with cancer urgently need new evidence of more personalized clinical and demographic characteristics and more sophisticated treatment and prevention strategies. However, no research has systematically analyzed the application and significance of artificial intelligence based on electronic health records in cancer care.

Objective

The aim of this study was to conduct a review to introduce the current state and limitations of artificial intelligence based on electronic health records of patients with cancer and to summarize the performance of artificial intelligence in mining electronic health records and its impact on cancer care.

Methods

Three databases were systematically searched to retrieve potentially relevant papers published from January 2009 to October 2020. Four principal reviewers assessed the quality of the papers and reviewed them for eligibility based on the inclusion criteria in the extracted data. The summary measures used in this analysis were the number and frequency of occurrence of the themes.

Results

Of the 1034 papers considered, 148 papers met the inclusion criteria. Cancer care, especially cancers of female organs and digestive organs, could benefit from artificial intelligence based on electronic health records through cancer emergencies and prognostic estimates, cancer diagnosis and prediction, tumor stage detection, cancer case detection, and treatment pattern recognition. The models can always achieve an area under the curve of 0.7. Ensemble methods and deep learning are on the rise. In addition, electronic medical records in the existing studies are mainly in English and from private institutional databases.

Conclusions

Artificial intelligence based on electronic health records performed well and could be useful for cancer care. Improving the performance of artificial intelligence can help patients receive more scientific-based and accurate treatments. There is a need for the development of new methods and electronic health record data sharing and for increased passion and support from cancer specialists.

Time to Pregnancy, Obstetrical and Neonatal Outcomes after Breast Cancer: A Study from the Maternity Network for Young Breast Cancer Patients

Although an increasing number of young breast cancer (BC) patients have a pregnancy desire after BC, the time necessary to obtain a pregnancy after treatment and subsequent outcomes remain unknown. We aimed to determine the time to evolutive pregnancy in a cohort of BC survivors and subsequent obstetrical and neonatal outcomes. We analyzed BC patients treated at Institut Curie from 2005-2017, aged 18-43 years old (y.o.) at diagnosis having at least one subsequent pregnancy. 133 patients were included, representing 197 pregnancies. Mean age at BC diagnosis was 32.8 y.o. and at pregnancy beginning was 36.8 y.o. 71% pregnancies were planned, 18% unplanned and 86% spontaneous. 64% pregnancies resulted in live birth (n = 131). Median time from BC diagnosis to pregnancy beginning was 48 months and was significantly associated with endocrine therapy (p < 0.001). Median time to pregnancy was 4.3 months. Median time to evolutive pregnancy 5.6 months. In multivariate analysis, menstrual cycles before pregnancy remained significantly associated with time to pregnancy and endocrine therapy with time evolutive to pregnancy. None of the BC treatments (chemotherapy/endocrine therapy/trastuzumab) was significantly associated with obstetrical nor neonatal outcomes, that seemed comparable to global population. Our findings provide reassuring data for pregnancy counseling both in terms of delay and outcome.