Establishing the role of honest broker: bridging the gap between protecting personal health data and clinical research efficiency

Classifying early infant feeding status from clinical notes using natural language processing and machine learning

The objective of this study is to develop and evaluate natural language processing (NLP) and machine learning models to predict infant feeding status from clinical notes in the Epic electronic health records system. The primary outcome was the classification of infant feeding status from clinical notes using Medical Subject Headings (MeSH) terms. Annotation of notes was completed using TeamTat to uniquely classify clinical notes according to infant feeding status. We trained 6 machine learning models to classify infant feeding status: logistic regression, random forest, XGBoost gradient descent, k-nearest neighbors, and support-vector classifier. Model comparison was evaluated based on overall accuracy, precision, recall, and F1 score. Our modeling corpus included an even number of clinical notes that was a balanced sample across each class. We manually reviewed 999 notes that represented 746 mother-infant dyads with a mean gestational age of 38.9 weeks and a mean maternal age of 26.6 years. The most frequent feeding status classification present for this study was exclusive breastfeeding [n = 183 (18.3%)], followed by exclusive formula bottle feeding [n = 146 (14.6%)], and exclusive feeding of expressed mother's milk [n = 102 (10.2%)], with mixed feeding being the least frequent [n = 23 (2.3%)]. Our final analysis evaluated the classification of clinical notes as breast, formula/bottle, and missing. The machine learning models were trained on these three classes after performing balancing and down sampling. The XGBoost model outperformed all others by achieving an accuracy of 90.1%, a macro-averaged precision of 90.3%, a macro-averaged recall of 90.1%, and a macro-averaged F1 score of 90.1%. Our results demonstrate that natural language processing can be applied to clinical notes stored in the electronic health records to classify infant feeding status. Early identification of breastfeeding status using NLP on unstructured electronic health records data can be used to inform precision public health interventions focused on improving lactation support for postpartum patients.

Factors Influencing Medication Administration Outcomes Among New Graduate Nurses Using Bar Code-Assisted Medication Administration

Paramount to patient safety is the ability for nurses to make clinical decisions free from human error. Yet, the dynamic clinical environment in which nurses work is characterized by uncertainty, urgency, and high consequence, necessitating that nurses make quick and critical decisions. The aim of this study was to examine the influence of human and environmental factors on the decision to administer among new graduate nurses in response to alert generation during bar code-assisted medication administration. The design for this study was a descriptive, longitudinal, observational cohort design using EHR audit log and administrative data. The study was set at a large, urban medical center in the United States and included 132 new graduate nurses who worked on adult, inpatient units. Research variables included human and environmental factors. Data analysis included descriptive and inferential analyses. This study found that participants continued with administration of a medication in 90.75% of alert encounters. When considering the response to an alert, residency cohort, alert category, and previous exposure variables were associated with the decision to proceed with administration. It is important to continue to study factors that influence nurses' decision-making, particularly during the process of medication administration, to improve patient safety and outcomes.

Describing Medication Administration and Alert Patterns Experienced by New Graduate Nurses During the First Year of Practice

The aim of this study was to describe medication administration and alert patterns among a cohort of new graduate nurses over the first year of practice. Medical errors related to clinical decision-making, including medication administration errors, may occur more frequently among new graduate nurses. To better understand nursing workflow and documentation workload in today's clinical environment, it is important to understand patterns of medication administration and alert generation during barcode-assisted medication administration. Study objectives were addressed through a descriptive, longitudinal, observational cohort design using secondary data analysis. Set in a large, urban medical center in the United States, the study sample included 132 new graduate nurses who worked on adult, inpatient units and administered medication using barcode-assisted medication administration. Data were collected through electronic health record and administration sources. New graduate nurses in the sample experienced a total of 587 879 alert and medication administration encounters, administering 772 unique medications to 17 388 unique patients. Nurses experienced an average medication workload of 28.09 medications per shift, 3.98% of which were associated with alerts, over their first year of practice. In addition to high volume of medication administration, new graduate nurses administer many different types of medications and are exposed to numerous alerts while using barcode-assisted medication administration.

Technology Platforms and Approaches for Building and Evaluating Machine Learning Methods in Healthcare

BACKGROUND

Artificial intelligence (AI) methods are becoming increasingly commonly implemented in healthcare as decision support, business intelligence tools, or, in some cases, Food and Drug Administration-approved clinical decision-makers. Advanced lab-based diagnostic tools are increasingly becoming AI driven. The path from data to machine learning methods is an active area for research and quality improvement, and there are few established best practices. With data being generated at an unprecedented rate, there is a need for processes that enable data science investigation that protect patient privacy and minimize other business risks. New approaches for data sharing are being utilized that lower these risks.

CONTENT

In this short review, clinical and translational AI governance is introduced along with approaches for securely building, sharing, and validating accurate and fair models. This is a constantly evolving field, and there is much interest in collecting data using standards, sharing data, building new models, evaluating models, sharing models, and, of course, implementing models into practice.

SUMMARY

AI is an active area of research and development broadly for healthcare and laboratory testing. Robust data governance and machine learning methodological governance are required. New approaches for data sharing are enabling the development of models and their evaluation. Evaluation of methods is difficult, particularly when the evaluation is performed by the team developing the method, and should ideally be prospective. New technologies have enabled standardization of platforms for moving analytics and data science methods.

Operating room time as a limited resource: ethical considerations for allocation

Scheduling surgical procedures among operating rooms (ORs) is mistakenly regarded as merely a tedious administrative task. However, the growing demand for surgical care and finite hours in a day qualify OR time as a limited resource. Accordingly, the objective of this manuscript is to reframe the process of OR scheduling as an ethical dilemma of allocating scarce medical resources. Recommendations for ethical allocation of OR time-based on both familiar and novel ethical values-are provided for healthcare institutions and individual surgeons.

Demographics, Changes in Treatment Patterns, and Outcomes of Bone and Soft Tissue Sarcomas in Korea-A Sarcoma-Specific, Institutional Registry-Based Analysis

Purpose

Because of the heterogeneity of sarcomas, establishing a well-collected, sarcoma-specific database is important for sarcoma research. We analyzed the first histology-based, sarcoma-specific institutional registry in Korea, which collected 28 years of patient data according to a predefined data format.

Patients and Methods

Adult bone and soft tissue sarcoma patients who were treated from June 1989 to January 2017 were identified and analyzed, based on the ICD-O-3 codes.

Results

Among the 3420 patients included, soft tissue and bone sarcomas comprised 77.8% (n = 2661) and 22.2% (n = 759), respectively. Median age at diagnosis was 50 (range, 16-98) in soft tissue sarcomas and 37 (range, 16-85) in bone sarcomas. Males and females comprised 45.5% and 54.5% of soft tissue sarcomas and 52.7% and 47.3% of bone sarcomas, respectively. Among the 3407 patients with treatment data available, 90.5% of the patients with soft tissue sarcomas and 80.8% of the patients with bone sarcomas received surgery first, of which 57.8% and 71.7% did not receive any subsequent treatment, respectively. Overall, the proportion of patients who received surgery alone decreased from 85.7% to 60.5% from the pre-2000 period to the 2010-2017 period. However, the use of adjuvant chemotherapy increased in patients with soft tissue sarcomas (from 8.0% to 17.2% in the same period), and the use of perioperative radiotherapy also increased in both groups (from 1.4% to 22.7% in soft tissue sarcomas, and 0% to 14.5% in bone sarcomas in the same period). In both soft tissue and bone sarcomas, old age (≥65 years) and diagnosis in the early study period were associated with poorer survival.

Conclusion

We presented a comprehensive summary of our sarcoma registry, including the demographics, changes in treatment patterns, and survival outcomes. This study will provide a framework for future studies.

Data Pseudonymization in a Range That Does Not Affect Data Quality: Correlation with the Degree of Participation of Clinicians

Personal medical information is an essential resource for research; however, there are laws that regulate its use, and it typically has to be pseudonymized or anonymized. When data are anonymized, the quantity and quality of extractable information decrease significantly. From the perspective of a clinical researcher, a method of achieving pseudonymized data without degrading data quality while also preventing data loss is proposed herein. As the level of pseudonymization varies according to the research purpose, the pseudonymization method applied should be carefully chosen. Therefore, the active participation of clinicians is crucial to transform the data according to the research purpose. This can contribute to data security by simply transforming the data through secondary data processing. Case studies demonstrated that, compared with the initial baseline data, there was a clinically significant difference in the number of datapoints added with the participation of a clinician (from 267,979 to 280,127 points, P < 0.001). Thus, depending on the degree of clinician participation, data anonymization may not affect data quality and quantity, and proper data quality management along with data security are emphasized. Although the pseudonymization level and clinical use of data have a trade-off relationship, it is possible to create pseudonymized data while maintaining the data quality required for a given research purpose. Therefore, rather than relying solely on security guidelines, the active participation of clinicians is important.

Measuring the immediate impact when first year pharmacy students are introduced to diverse career pathways

BACKGROUND AND PURPOSE

To assess if exposure to diverse pharmacy career pathways influences the Doctor of Pharmacy (PharmD) student's career plans within the first month of an academic curriculum.

EDUCATIONAL ACTIVITY AND SETTING

First year PharmD students were enrolled in a four-week course with a focus on introduction to core practice areas of pharmacy: community, hospital, and managed care. Guidance was provided with resources and a pharmacist panel to aid in both self-learning and direct sharing about diverse areas of pharmacy practice extending beyond the core course practice areas. A survey was given at the beginning and at the end of the course to measure the influence of course activities on first year students' aspirations for varied pharmacy careers. All students completed the survey but needed to opt into the research for data to be collected. Chi-square, Fisher's Exact Test, and descriptive statistics were used in data analysis.

FINDINGS

In this study of 508 first year pharmacy students, we found that 50.8% reported a change in their pharmacy career plans at the end of the course. Student interest in non-traditional career paths increased from 38.2% at the beginning of the course to 47.6% at the end of the course. As a result, students reported that they would select different electives (P < .001), pursue different pharmacy organizations (P = .0003), and explore new internship opportunities (P < .001). Overall, 98% found the course introduced them to pharmacy career paths they were previously unaware existed.

SUMMARY

Early exposure to diverse pharmacy career pathways influences students' career plans.

Data Sharing of Imaging in an Evolving Health Care World: Report of the ACR Data Sharing Workgroup Part 2: Annotation, Curation, and Contracting

A core principle of ethical data sharing is maintaining the security and anonymity of the data, and care must be taken to ensure medical records and images cannot be reidentified to be traced back to patients or misconstrued as a breach in the trust between health care providers and patients. Once those principles have been observed, those seeking to share data must take the appropriate steps to curate the data in a way that organizes the clinically relevant information so as to be useful to the data sharing party, assesses the ensuing value of the data set and its annotations, and informs the data sharing contracts that will govern use of the data. Embarking on a data sharing partnership engenders a host of ethical, practical, technical, legal, and commercial challenges that require a thoughtful, considered approach. In 2019 the ACR convened a Data Sharing Workgroup to develop philosophies around best practices in the sharing of health information. This is Part 2 of a Report on the workgroup's efforts in exploring these issues.

Implementing Universal Suicide Risk Screening in a Pediatric Hospital

BACKGROUND

Health care providers are in a prime position to identify teens at risk for suicide, yet many do not. The research team developed and implemented a hospitalwide program to identify teens at elevated risk for suicide and connect them with services.

METHODS

Screening was implemented at both locations of a pediatric hospital, including two emergency departments, three urgent care clinics, and ambulatory clinics. Patients aged 12 years and older presenting for care were screened for suicide risk using the Ask Suicide-Screening Questions (ASQ) in most settings, while the Columbia-Suicide Severity Rating Scale (C-SSRS) was used in mental health areas. A social worker responded to positive screens to complete a more thorough assessment and determine next steps. Social workers also completed outreach to patients in the weeks following a positive screen. Implementation began with pilot locations and expanded after refinements were made. Stakeholders provided screening recommendations, and education was provided prior to implementation. The cost of implementation was calculated based on the time screening required from nursing and social work.

RESULTS

Review of the program focused on implementation fidelity, quality improvement, and trends among screening results. During the first year of screening, 138,598 screens were completed, and 6.8% of screens were positive for elevated risk. The annualized cost of the program was estimated to be $887,708.65 for personnel directly involved in screening and following up on positive screens.

CONCLUSION

Early involvement of stakeholders and hospital leaders and a robust response plan were essential to successful implementation of this suicide-screening program.

Development and maintenance of a medical education research registry

BACKGROUND

Medical Education research suffers from several methodological limitations including too many single institution, small sample-sized studies, limited access to quality data, and insufficient institutional support. Increasing calls for medical education outcome data and quality improvement research have highlighted a critical need for uniformly clean and easily accessible data. Research registries may fill this gap. In 2006, the Research on Medical Education Outcomes (ROMEO) unit of the Program for Medical Innovations and Research (PrMEIR) at New York University's (NYU) Robert I. Grossman School of Medicine established the Database for Research on Academic Medicine (DREAM). DREAM is a database of routinely collected, de-identified undergraduate (UME, medical school leading up to the Medical Doctor degree) and graduate medical education (GME, residency also known as post graduate education leading to eligibility for specialty board certification) outcomes data available, through application, to researchers. Learners are added to our database through annual consent sessions conducted at the start of educational training. Based on experience, we describe our methods in creating and maintaining DREAM to serve as a guide for institutions looking to build a new or scale up their medical education registry.

RESULTS

At present, our UME and GME registries have consent rates of 90% (n = 1438/1598) and 76% (n = 1988/2627), respectively, with a combined rate of 81% (n = 3426/4225). 7% (n = 250/3426) of these learners completed both medical school and residency at our institution. DREAM has yielded a total of 61 individual studies conducted by medical education researchers and a total of 45 academic journal publications.

CONCLUSION

We have built a community of practice through the building of DREAM and hope, by persisting in this work the full potential of this tool and the community will be realized. While researchers with access to the registry have focused primarily on curricular/ program evaluation, learner competency assessment, and measure validation, we hope to expand the output of the registry to include patient outcomes by linking learner educational and clinical performance across the UME-GME continuum and into independent practice. Future publications will reflect our efforts in reaching this goal and will highlight the long-term impact of our collaborative work.

Next-Generation Sequencing-Based Cancer Panel Data Conversion Using International Standards to Implement a Clinical Next-Generation Sequencing Research System: Single-Institution Study

BACKGROUND

The analytical capacity and speed of next-generation sequencing (NGS) technology have been improved. Many genetic variants associated with various diseases have been discovered using NGS. Therefore, applying NGS to clinical practice results in precision or personalized medicine. However, as clinical sequencing reports in electronic health records (EHRs) are not structured according to recommended standards, clinical decision support systems have not been fully utilized. In addition, integrating genomic data with clinical data for translational research remains a great challenge.

OBJECTIVE

To apply international standards to clinical sequencing reports and to develop a clinical research information system to integrate standardized genomic data with clinical data.

METHODS

We applied the recently published ISO/TS 20428 standard to 367 clinical sequencing reports generated by panel (91 genes) sequencing in EHRs and implemented a clinical NGS research system by extending the clinical data warehouse to integrate the necessary clinical data for each patient. We also developed a user interface with a clinical research portal and an NGS result viewer.

RESULTS

A single clinical sequencing report with 28 items was restructured into four database tables and 49 entities. As a result, 367 patients' clinical sequencing data were connected with clinical data in EHRs, such as diagnosis, surgery, and death information. This system can support the development of cohort or case-control datasets as well.

CONCLUSIONS

The standardized clinical sequencing data are not only for clinical practice and could be further applied to translational research.

Tools and Methods for Real-World Evidence Generation: Pragmatic Trials, Electronic Consent, and Data Linkages

Real-world evidence requires use of new tools and methods to support efficient evidence generation. Among those tools are pragmatic trials, utilization of central/single institutional review board and electronic consent, and data linkages between diverse types of data sources (eg, a trial or registry to administrative claims or electronic medical record data). This article reviews these topics in the context of describing several exemplar use cases specific to rheumatology and provides perspective regarding both the promise and potential pitfalls in using these tools and approaches.

Use and Understanding of Anonymization and De-Identification in the Biomedical Literature: Scoping Review

Background

The secondary use of health data is central to biomedical research in the era of data science and precision medicine. National and international initiatives, such as the Global Open Findable, Accessible, Interoperable, and Reusable (GO FAIR) initiative, are supporting this approach in different ways (eg, making the sharing of research data mandatory or improving the legal and ethical frameworks). Preserving patients’ privacy is crucial in this context. De-identification and anonymization are the two most common terms used to refer to the technical approaches that protect privacy and facilitate the secondary use of health data. However, it is difficult to find a consensus on the definitions of the concepts or on the reliability of the techniques used to apply them. A comprehensive review is needed to better understand the domain, its capabilities, its challenges, and the ratio of risk between the data subjects’ privacy on one side, and the benefit of scientific advances on the other.

Objective

This work aims at better understanding how the research community comprehends and defines the concepts of de-identification and anonymization. A rich overview should also provide insights into the use and reliability of the methods. Six aspects will be studied: (1) terminology and definitions, (2) backgrounds and places of work of the researchers, (3) reasons for anonymizing or de-identifying health data, (4) limitations of the techniques, (5) legal and ethical aspects, and (6) recommendations of the researchers.

Methods

Based on a scoping review protocol designed a priori, MEDLINE was searched for publications discussing de-identification or anonymization and published between 2007 and 2017. The search was restricted to MEDLINE to focus on the life sciences community. The screening process was performed by two reviewers independently.

Results

After searching 7972 records that matched at least one search term, 135 publications were screened and 60 full-text articles were included. (1) Terminology: Definitions of the terms de-identification and anonymization were provided in less than half of the articles (29/60, 48%). When both terms were used (41/60, 68%), their meanings divided the authors into two equal groups (19/60, 32%, each) with opposed views. The remaining articles (3/60, 5%) were equivocal. (2) Backgrounds and locations: Research groups were based predominantly in North America (31/60, 52%) and in the European Union (22/60, 37%). The authors came from 19 different domains; computer science (91/248, 36.7%), biomedical informatics (47/248, 19.0%), and medicine (38/248, 15.3%) were the most prevalent ones. (3) Purpose: The main reason declared for applying these techniques is to facilitate biomedical research. (4) Limitations: Progress is made on specific techniques but, overall, limitations remain numerous. (5) Legal and ethical aspects: Differences exist between nations in the definitions, approaches, and legal practices. (6) Recommendations: The combination of organizational, legal, ethical, and technical approaches is necessary to protect health data.

Conclusions

Interest is growing for privacy-enhancing techniques in the life sciences community. This interest crosses scientific boundaries, involving primarily computer science, biomedical informatics, and medicine. The variability observed in the use of the terms de-identification and anonymization emphasizes the need for clearer definitions as well as for better education and dissemination of information on the subject. The same observation applies to the methods. Several legislations, such as the American Health Insurance Portability and Accountability Act (HIPAA) and the European General Data Protection Regulation (GDPR), regulate the domain. Using the definitions they provide could help address the variable use of these two concepts in the research community.

Personal health data: A systematic mapping study

BACKGROUND

Personal health data (PHD) research has been intensified over the last years, attracting the attention of scientists from different fields, such as software engineers, computer scientists and medical professionals. The increasing interest of researchers can be attributed to the exponential growth of the available PHD due to the widespread adoption of ubiquitous technology in everyday life, as well as to the potential of the ongoing digital transformation in healthcare. This increasing interest requires that academia has an overview of the published scientific literature to plan future endeavors.

OBJECTIVE

The main objective of this study is to identify and address research gaps in literature regarding PHD.

METHOD

This paper conducts a systematic mapping study to summarize the existing PHD approaches in literature and to organize the selected studies according to six classification criteria: publication source, publication year, research types, empirical types, contribution types and research topic.

RESULTS

In total 79 papers have been included after fulfilling the inclusion criteria and have been classified accordingly. There is an increasing amount of attention that has been paid to PHD since 2014. The majority of papers is published in journals. The two main research types found were solution proposals and evaluation research. The majority of the selected papers were empirically evaluated. The main contribution types were methods and frameworks. Data privacy is the most frequently addressed topic in PHD literature, followed by data sharing.

CONCLUSIONS

The findings of this systematic mapping study have implications for both researchers who are planning new studies in PHD and for practitioners who are working in connected health and would like to have an overview on the existent studies on PHD research area.

Using Time-Referenced Data to Assess Medication Administration Performance and Quality

OBJECTIVE

This study tests the feasibility of using a large (big) clinical data set to test the ability to extract time-referenced data related to medication administration to identify late doses and as-needed (PRN) administration patterns by RNs in an inpatient setting.

METHODS

The study is a secondary analysis of a set of data using bar-code medication administration time stamps (n = 3043812) for 50883 patients admitted to a single, urban, 525-bed hospital in 11 inpatient units by 714 nurses between April 1, 2013, and March 31, 2015.

RESULTS

The large majority of scheduled medications (43.3%) were administered between 9 to 10 AM and 9 to 10 PM accounting for the most amount of delayed doses. On average, patients received 8.9 medications per day, and nurses administered 19.7 medications per shift. The average full-time nurse administered 3414 medications per year.

CONCLUSIONS

The findings support use of time-referenced data to identify clinical processes and performance in administering scheduled and PRN medications.