Exploring the tradeoff between data privacy and utility with a clinical data analysis use case.

BACKGROUND: Securing adequate data privacy is critical for the productive utilization of data. De-identification, involving masking or replacing specific values in a dataset, could damage the dataset's utility. However, finding a reasonable balance between data privacy and utility is not straightforward. Nonetheless, few studies investigated how data de-identification efforts affect data analysis results. This study aimed to demonstrate the effect of different de-identification methods on a dataset's utility with a clinical analytic use case and assess the feasibility of finding a workable tradeoff between data privacy and utility. METHODS: Predictive modeling of emergency department length of stay was used as a data analysis use case. A logistic regression model was developed with 1155 patient cases extracted from a clinical data warehouse of an academic medical center located in Seoul, South Korea. Nineteen de-identified datasets were generated based on various de-identification configurations using ARX, an open-source software for anonymizing sensitive personal data. The variable distributions and prediction results were compared between the de-identified datasets and the original dataset. We examined the association between data privacy and utility to determine whether it is feasible to identify a viable tradeoff between the two. RESULTS: All 19 de-identification scenarios significantly decreased re-identification risk. Nevertheless, the de-identification processes resulted in record suppression and complete masking of variables used as predictors, thereby compromising dataset utility. A significant correlation was observed only between the re-identification reduction rates and the ARX utility scores. CONCLUSIONS: As the importance of health data analysis increases, so does the need for effective privacy protection methods. While existing guidelines provide a basis for de-identifying datasets, achieving a balance between high privacy and utility is a complex task that requires understanding the data's intended use and involving input from data users. This approach could help find a suitable compromise between data privacy and utility.

Predicting Workplace Violence in the Emergency Department Based on Electronic Health Record Data.

INTRODUCTION: Emergency departments are extremely vulnerable to workplace violence, and emergency nurses are frequently exposed to workplace violence. We developed workplace violence prediction models using machine learning methods based on data from electronic health records. METHODS: This study was conducted using electronic health record data collected between January 1, 2016 and December 31, 2021. Workplace violence cases were identified based on violence-related mentions in nursing records. Workplace violence was predicted using various factors related to emergency department visit and stay. RESULTS: The dataset included 1215 workplace violence cases and 6044 nonviolence cases. Random Forest showed the best performance among the algorithms adopted in this study. Workplace violence was predicted with higher accuracy when both ED visit and ED stay factors were used as predictors (0.90, 95% confidence interval 0.898-0.912) than when only ED visit factors were used. When both ED visit and ED stay factors were included for prediction, the strongest predictor of risk of WPV was patient dissatisfaction, followed by high average daily length of stay, high daily number of patients, and symptoms of psychiatric disorders. DISCUSSION: This study showed that workplace violence could be predicted with previous data regarding ED visits and stays documented in electronic health records. Timely prediction and mitigation of workplace violence could improve the safety of emergency nurses and the quality of nursing care. To prevent workplace violence, emergency nurses must recognize and continuously observe the risk factors for workplace violence from admission to discharge.

The Contribution of Material, Behavioral, Psychological, and Social-Relational Factors to Income-Related Disparities in Cardiovascular Risk Among Older Adults.

BACKGROUND: Understanding the factors underlying health disparities is vital to developing strategies to improve health equity in old age. Such efforts should be encouraged in Korea. OBJECTIVE: This study explored how material, behavioral, psychological, and social-relational factors contribute to income-related disparities in cardiovascular risk among Korean adults 65 years and older. METHODS: This was a secondary analysis of Korean National Health and Nutrition Examination Survey data (2013-2017), targeting 7347 older adults (≥65 years). Socioeconomic position, defined as income, was the primary indicator. The outcome was binary for predicted cardiovascular risk (<90 vs ≥90 percentile). Disparities were measured using relative index of inequality (RII). The contributions of material, behavioral, psychological, and social-relational factors were estimated by calculating percentage reduction in RII when adjusted for these factors. RESULTS: Among men aged 65 to 74 years and women 75 years or older, the largest reductions in RII were achieved after adjusting for social-relational factors. Among men 75 years or older and women aged 65 to 74 years, adjusting for material factors resulted in the largest reductions in RII. Adjustments for behavioral factors also reduced RII for both genders aged 65 to 74 years. CONCLUSIONS: Improving the social, material, and behavioral circumstances of lower-income older adults may help address income-related disparities in cardiovascular risk in old age.

A technology-based information and coaching/support program and self-efficacy of Asian American breast cancer survivors.

Women's self-efficacy for coping with breast cancer is one of the key factors that lead to successful breast cancer survivorship. Due to the cultural stigma linked to breast cancer (e.g., breast cancer is a genetic disease), Asian Americans are known as a high-risk group within breast cancer survivors. However, healthcare providers are challenged to promote women's self-efficacy while considering their cultural beliefs and attitudes. In this study, the efficacy of a technology-based information and coaching/support program was examined in improving self-efficacy for coping with breast cancer among Asian American survivors. A randomized repeated measures control group study was conducted with 67 Asian American breast cancer survivors. The questions on background characteristics, the Personal Resource Questionnaire, the Perceived Isolation Scale, the Supportive Care Needs Survey Short Form 34, and the Cancer Behavior Inventory were used. The data were analyzed using repeated measurement analyses, χ2 tests, and decision tree analyses. There were significant increases in the self-efficacy scores of both control and intervention groups over time (p = .017). However, the increase in the control group's self-efficacy scores was only up to post 1 month, and there was a decrease in the scores by post 3 months. When the participants were divided into high and low-change groups based on the changes in their self-efficacy scores for 3 months, the intervention group had more participants who belonged to the high-change group (p = .036). The technology-based intervention was effective in improving self-efficacy for coping with breast cancer among Asian American breast cancer survivors.

Representing Nursing Data With Fast Healthcare Interoperability Resources: Early Lessons Learned With a Use Case Scenario on Home-Based Pressure Ulcer Care.

Healthcare communities are rapidly embracing Health Level 7's Fast Healthcare Interoperability Resources standard as the next-generation messaging protocol to facilitate data interoperability. Implementation-friendly formats for data representation and compliance to widely adopted industry standards are among the strengths of Fast Healthcare Interoperability Resources that are accelerating its wide adoption. Research confirms the advantages of Fast Healthcare Interoperability Resources in increasing data interoperability in mortality reporting, genetic test sharing, and patient-generated data. However, few studies have investigated the application of Fast Healthcare Interoperability Resources in nursing-specific domains. In this study, a Fast Healthcare Interoperability Resources document was generated for a use case scenario in a home-based, pressure ulcer care setting. Study goals were to describe the step-by-step process of generating a Fast Healthcare Interoperability Resources artifact and to inform nursing communities about the advantages and challenges in representing nursing data with Fast Healthcare Interoperability Resources. Overall, Fast Healthcare Interoperability Resources effectively represented the majority of the data included in the use case scenario. A few challenges that could potentially cause information loss were noted such as the lack of standardized concept codes for value encoding and the difficulty directly connecting an observation to a related condition. Continuous evaluations in diverse nursing domains are needed in order to gain a more thorough insight on potential challenges that Fast Healthcare Interoperability Resources holds in representing nursing data.

Optimization of Nursing-Specific Flu Alerts.

Clinical decision support interventions, such as alerts and reminders, can improve clinician compliance with practice guidelines and patient outcomes. Alerts that trigger at inappropriate times are often dismissed by clinicians, reducing desired actions rather than increasing them. A set of nursing-specific alerts related to influenza screening and vaccination were optimized so that they would "trigger" less often but function adequately to maintain institutional flu vaccination compliance. We analyzed the current triggering criteria for six flu vaccine-related alerts and asked nurse end users for suggestions to increase specificity. Using the "five rights" (of clinical decision support) as a framework, alerts were redesigned to address user needs. New alerts were tested and implemented and their activity compared in two different flu seasons, preoptimization and postoptimization. The redesigned alerts resulted in fewer alerts per encounter (P < .0001), less dismissals of alerts (P < .0001), and a 2.8% point improvement in compliance rates for flu vaccine screening, documentation, and administration. A focus group confirmed that the redesign improved workflow, but some nurses thought they still triggered too often. The five rights model can support improvements in alert design and outcomes.

Developing a Physical Activity Ontology to Support the Interoperability of Physical Activity Data.

BACKGROUND: Physical activity data provides important information on disease onset, progression, and treatment outcomes. Although analyzing physical activity data in conjunction with other clinical and microbiological data will lead to new insights crucial for improving human health, it has been hampered partly because of the large variations in the way the data are collected and presented. OBJECTIVE: The aim of this study was to develop a Physical Activity Ontology (PACO) to support structuring and standardizing heterogeneous descriptions of physical activities. METHODS: We prepared a corpus of 1140 unique sentences collected from various physical activity questionnaires and scales as well as existing standardized terminologies and ontologies. We extracted concepts relevant to physical activity from the corpus using a natural language processing toolkit called Multipurpose Text Processing Tool. The target concepts were formalized into an ontology using Protégé (version 4). Evaluation of PACO was performed to ensure logical and structural consistency as well as adherence to the best practice principles of building an ontology. A use case application of PACO was demonstrated by structuring and standardizing 36 exercise habit statements and then automatically classifying them to a defined class of either sufficiently active or insufficiently active using FaCT++, an ontology reasoner available in Protégé. RESULTS: PACO was constructed using 268 unique concepts extracted from the questionnaires and assessment scales. PACO contains 225 classes including 9 defined classes, 20 object properties, 1 data property, and 23 instances (excluding 36 exercise statements). The maximum depth of classes is 4, and the maximum number of siblings is 38. The evaluations with ontology auditing tools confirmed that PACO is structurally and logically consistent and satisfies the majority of the best practice rules of ontology authoring. We showed in a small sample of 36 exercise habit statements that we could formally represent them using PACO concepts and object properties. The formal representation was used to infer a patient activity status category of sufficiently active or insufficiently active using the FaCT++ reasoner. CONCLUSIONS: As a first step toward standardizing and structuring heterogeneous descriptions of physical activities for integrative data analyses, PACO was constructed based on the concepts collected from physical activity questionnaires and assessment scales. PACO was evaluated to be structurally consistent and compliant to ontology authoring principles. PACO was also demonstrated to be potentially useful in standardizing heterogeneous physical activity descriptions and classifying them into clinically meaningful categories that reflect adequacy of exercise.

Deep mining heterogeneous networks of biomedical linked data to predict novel drug-target associations.

MOTIVATION: A heterogeneous network topology possessing abundant interactions between biomedical entities has yet to be utilized in similarity-based methods for predicting drug-target associations based on the array of varying features of drugs and their targets. Deep learning reveals features of vertices of a large network that can be adapted in accommodating the similarity-based solutions to provide a flexible method of drug-target prediction. RESULTS: We propose a similarity-based drug-target prediction method that enhances existing association discovery methods by using a topology-based similarity measure. DeepWalk, a deep learning method, is adopted in this study to calculate the similarities within Linked Tripartite Network (LTN), a heterogeneous network generated from biomedical linked datasets. This proposed method shows promising results for drug-target association prediction: 98.96% AUC ROC score with a 10-fold cross-validation and 99.25% AUC ROC score with a Monte Carlo cross-validation with LTN. By utilizing DeepWalk, we demonstrate that: (i) this method outperforms other existing topology-based similarity computation methods, (ii) the performance is better for tripartite than with bipartite networks and (iii) the measure of similarity using network topology outperforms the ones derived from chemical structure (drugs) or genomic sequence (targets). Our proposed methodology proves to be capable of providing a promising solution for drug-target prediction based on topological similarity with a heterogeneous network, and may be readily re-purposed and adapted in the existing of similarity-based methodologies. AVAILABILITY AND IMPLEMENTATION: The proposed method has been developed in JAVA and it is available, along with the data at the following URL: https://github.com/zongnansu1982/drug-target-prediction . CONTACT: nazong@ucsd.edu. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Selecting Optimal Subset to release under Differentially Private M-estimators from Hybrid Datasets.

Privacy concern in data sharing especially for health data gains particularly increasing attention nowadays. Now some patients agree to open their information for research use, which gives rise to a new question of how to effectively use the public information to better understand the private dataset without breaching privacy. In this paper, we specialize this question as selecting an optimal subset of the public dataset for M-estimators in the framework of differential privacy (DP) in [1]. From a perspective of non-interactive learning, we first construct the weighted private density estimation from the hybrid datasets under DP. Along the same line as [2], we analyze the accuracy of the DP M-estimators based on the hybrid datasets. Our main contributions are (i) we find that the bias-variance tradeoff in the performance of our M-estimators can be characterized in the sample size of the released dataset; (2) based on this finding, we develop an algorithm to select the optimal subset of the public dataset to release under DP. Our simulation studies and application to the real datasets confirm our findings and set a guideline in the real application.

PATTERN: Pain Assessment for paTients who can't TEll using Restricted Boltzmann machiNe.

BACKGROUND: Accurately assessing pain for those who cannot make self-report of pain, such as minimally responsive or severely brain-injured patients, is challenging. In this paper, we attempted to address this challenge by answering the following questions: (1) if the pain has dependency structures in electronic signals and if so, (2) how to apply this pattern in predicting the state of pain. To this end, we have been investigating and comparing the performance of several machine learning techniques. METHODS: We first adopted different strategies, in which the collected original n-dimensional numerical data were converted into binary data. Pain states are represented in binary format and bound with above binary features to construct (n + 1) -dimensional data. We then modeled the joint distribution over all variables in this data using the Restricted Boltzmann Machine (RBM). RESULTS: Seventy-eight pain data items were collected. Four individuals with the number of recorded labels larger than 1000 were used in the experiment. Number of avaliable data items for the four patients varied from 22 to 28. Discriminant RBM achieved better accuracy in all four experiments. CONCLUSION: The experimental results show that RBM models the distribution of our binary pain data well. We showed that discriminant RBM can be used in a classification task, and the initial result is advantageous over other classifiers such as support vector machine (SVM) using PCA representation and the LDA discriminant method.

Educational Needs Assessment for Family Caregivers of Dementia Patients.

People living with dementia are highly dependent on caregivers. We conducted an online survey with regard to caregivers' educational experiences, needs, and expectations. We found that most of the participants lacked educational experiences and expected updated methods through metaverse in virtual reality. Therefore, future studies should verify the effectiveness of education.

Quality of Person-Generated Healthy Walking Data: An Explorative Analysis.

Despite the potential benefits of Person Generated Health Data (PGHD), data quality issues impede its use. This study examined the effect of different methods for filtering armband data on determining the amount of healthy walking and the consistency between healthy walking captured using armbands and health diaries. Four weeks of armband and health diary data were acquired from 103 college students. Armband data filtering was performed using heart rate measures and minimum daily step counts as a proxy for adequate daily wear time. No substantial differences in the filtered armband datasets were observed by filtering methods. Significant gaps were observed between healthy walking amounts determined from armband data and through the health diary. Future studies need to explore more diverse data filtering methods and their impact on health outcome assessments.

Evaluating the validity of the nursing statements algorithmically generated based on the International Classifications of Nursing Practice for respiratory nursing care using large language models.

OBJECTIVE: This study aims to facilitate the creation of quality standardized nursing statements in South Korea's hospitals using algorithmic generation based on the International Classifications of Nursing Practice (ICNP) and evaluation through Large Language Models. MATERIALS AND METHODS: We algorithmically generated 15 972 statements related to acute respiratory care using 117 concepts and concept composition models of ICNP. Human reviewers, Generative Pre-trained Transformers 4.0 (GPT-4.0), and Bio_Clinical Bidirectional Encoder Representations from Transformers (BERT) evaluated the generated statements for validity. The evaluation by GPT-4.0 and Bio_ClinicalBERT was conducted with and without contextual information and training. RESULTS: Of the generated statements, 2207 were deemed valid by expert reviewers. GPT-4.0 showed a zero-shot  AUC of 0.857, which aggravated with contextual information. Bio_ClinicalBERT, after training, significantly improved, reaching an AUC of 0.998. CONCLUSION: Bio_ClinicalBERT effectively validates auto-generated nursing statements, offering a promising solution to enhance and streamline healthcare documentation processes.

Identifying appropriate reference data models for comparative effectiveness research (CER) studies based on data from clinical information systems.

INTRODUCTION: The need for a common format for electronic exchange of clinical data prompted federal endorsement of applicable standards. However, despite obvious similarities, a consensus standard has not yet been selected in the comparative effectiveness research (CER) community. METHODS: Using qualitative metrics for data retrieval and information loss across a variety of CER topic areas, we compare several existing models from a representative sample of organizations associated with clinical research: the Observational Medical Outcomes Partnership (OMOP), Biomedical Research Integrated Domain Group, the Clinical Data Interchange Standards Consortium, and the US Food and Drug Administration. RESULTS: While the models examined captured a majority of the data elements that are useful for CER studies, data elements related to insurance benefit design and plans were most detailed in OMOP's CDM version 4.0. Standardized vocabularies that facilitate semantic interoperability were included in the OMOP and US Food and Drug Administration Mini-Sentinel data models, but are left to the discretion of the end-user in Biomedical Research Integrated Domain Group and Analysis Data Model, limiting reuse opportunities. Among the challenges we encountered was the need to model data specific to a local setting. This was handled by extending the standard data models. DISCUSSION: We found that the Common Data Model from the OMOP met the broadest complement of CER objectives. Minimal information loss occurred in mapping data from institution-specific data warehouses onto the data models from the standards we assessed. However, to support certain scenarios, we found a need to enhance existing data dictionaries with local, institution-specific information.

Factors affecting the length of stay in the emergency department for critically Ill patients transferred to regional emergency medical center.

AIM: To identify the factors affecting Emergency Department Length of Stay for transferred critically ill patients. BACKGROUND: The Length of Stay of the transferred patients is an important indicator of Emergency Department service quality; thus, understanding the factors affecting the Emergency Department Length of Stay of transferred critically ill patients is essential. METHODS: Using the electronic medical records of 968 transferred critically ill Emergency Department patients of a tertiary hospital in Korea, prediction models for Emergency Department Length of Stay were built using various machine learning algorithms. RESULTS: The logistic regression (AUROC 0.85) models showed the best performance, followed by random forest (AUROC 0.83) and Naive Bayes (AUROC 0.83). The logistic regression model indicated that fewer consultations, the highest acuity level, need for an emergency operation or angiography, need for ICU admission, severe emergency disease and fewer diagnoses were the statistically significant predictors for Emergency Department Length of Stay of 6 h or less. CONCLUSIONS: The transferred critically ill patients analysed in this study who required immediate or specialized care tended to receive needed care on time at the study site. IMPLICATIONS FOR NURSING MANAGEMENT: Understanding the factors affecting the Emergency Department Length of Stay of transferred critically ill patients is crucial for developing strategies to manage the nursing resource of Emergency Department successfully.

From intention to action: The factors affecting health data sharing intention and action.

INTRODUCTION: Effective prevention and treatment of diseases requires utilization of health-related lifestyle data, which has thus become increasingly important. According to some studies, participants were willing to share their health data for use in medical care and research. Although intention does not always accurately reflect action, few studies have examined the question of whether data-sharing intention leads to data-sharing action. OBJECTIVE: The aim of this study was to examine the extent of actualizing data-sharing intention to data-sharing action and to identify the factors that influence data-sharing intention and action. METHODS: A web-based survey of members of a university examined the data-sharing intention and issues of concern when making decisions on data sharing. The participants were asked to deposit their armband data for use in research at the end of the survey. A comparison of data-sharing intention and action in relation to the participants' characteristics was performed. Factors having a significant effect on data-sharing intention and action were identified using logistic regressions. RESULTS: Of 386 participants, 294 expressed willingness to share health data. However, only 73 participants deposited their armband data. The primary reason for refusal to deposit armband data was the inconvenience of the data transfer process (56.3%). Appropriate compensation had a significant effect on data-sharing intention (OR: 3.3, CI: 1.86-5.75) and action (OR: 2.8, CI: 1.14-8.21). The compensation for data sharing (OR:2.8, CI:1.14-8.21) and familiarity with data (OR:3.1, CI:1.36-8.21) were significant predictors of data sharing action, however, data-sharing intention was not (OR: 1.5, CI:0.65-3.72). CONCLUSION: Despite expressing willingness to share their health data, the participants' intention was not actualized to data-sharing behavior for depositing armband data. Implementation of a streamlined data transfer process and providing appropriate compensation might facilitate data-sharing. These findings could be useful in development of strategies to facilitate sharing and reuse of health data.

Essential properties and explanation effectiveness of explainable artificial intelligence in healthcare: A systematic review.

Background: Significant advancements in the field of information technology have influenced the creation of trustworthy explainable artificial intelligence (XAI) in healthcare. Despite improved performance of XAI, XAI techniques have not yet been integrated into real-time patient care. Objective: The aim of this systematic review is to understand the trends and gaps in research on XAI through an assessment of the essential properties of XAI and an evaluation of explanation effectiveness in the healthcare field. Methods: A search of PubMed and Embase databases for relevant peer-reviewed articles on development of an XAI model using clinical data and evaluating explanation effectiveness published between January 1, 2011, and April 30, 2022, was conducted. All retrieved papers were screened independently by the two authors. Relevant papers were also reviewed for identification of the essential properties of XAI (e.g., stakeholders and objectives of XAI, quality of personalized explanations) and the measures of explanation effectiveness (e.g., mental model, user satisfaction, trust assessment, task performance, and correctability). Results: Six out of 882 articles met the criteria for eligibility. Artificial Intelligence (AI) users were the most frequently described stakeholders. XAI served various purposes, including evaluation, justification, improvement, and learning from AI. Evaluation of the quality of personalized explanations was based on fidelity, explanatory power, interpretability, and plausibility. User satisfaction was the most frequently used measure of explanation effectiveness, followed by trust assessment, correctability, and task performance. The methods of assessing these measures also varied. Conclusion: XAI research should address the lack of a comprehensive and agreed-upon framework for explaining XAI and standardized approaches for evaluating the effectiveness of the explanation that XAI provides to diverse AI stakeholders.

Attitudes and perspectives of nurses and physicians in South Korea towards the clinical use of person-generated health data.

This study aimed to explore the adoption of person-generated health data in clinical settings and discern the factors influencing clinicians' willingness to use it. A web-based survey containing 48 questions was developed based on prior research and the Unified Theory of Acceptance and Use of Technology 2 model. The survey was administered to a convenience sample of 486 nurses and physicians in South Korea recruited through an online community and snowball sampling. Of these, 70.7% were physicians. While 65% had used mobile health apps and devices, only 12.8% were familiar with person-generated health data. Still, a promising 73.3% expressed interest in incorporating person-generated health data into patient care, particularly data on blood glucose and vital signs. The findings of the study also indicated that clinicians specializing in internal medicine (OR: 1.9, CI: 1.16-3.19), familiar with person-generated health data (OR: 2.6, CI: 1.58-4.29), with a positive view of information and communication technology adoption (OR: 2.6, CI: 1.65-4.13), and who see the value in person-generated health data (OR: 3.9, CI: 2.55-6.09) showed higher inclination to utilize it. However, those in outpatient settings (OR: 0.4, CI: 0.19-0.73) showed less enthusiasm. The findings of this study suggest that despite the willingness of clinicians to use person-generated health data, various barriers must be addressed first, including a lack of knowledge regarding its use, concerns about data reliability and quality, and a lack of provider incentives. Overcoming these challenges demands concerted organizational or policy support. This research underscores person-generated health data's untapped potential in healthcare and the pressing need for strategies that facilitate its clinical integration.

An approach to improve LOINC mapping through augmentation of local test names.

Mapping medical test names into a standardized vocabulary is a prerequisite to sharing test-related data between health care entities. One major barrier in this process is the inability to describe tests in sufficient detail to assign the appropriate name in Logical Observation Identifiers, Names, and Codes (LOINC®). Approaches to address mapping of test names with incomplete information have not been well described. We developed a process of "enhancing" local test names by incorporating information required for LOINC mapping into the test names themselves. When using the Regenstrief LOINC Mapping Assistant (RELMA) we found that 73/198 (37%) of "enhanced" test names were successfully mapped to LOINC, compared to 41/191 (21%) of original names (p=0.001). Our approach led to a significantly higher proportion of test names with successful mapping to LOINC, but further efforts are required to achieve more satisfactory results.

A workflow-oriented framework-driven implementation and local adaptation of clinical information systems: a case study of nursing documentation system implementation at a tertiary rehabilitation hospital.

Health information systems are often designed and developed without integrating users' specific needs and preferences. This decreases the users' productivity, satisfaction, and acceptance of the system and increases the necessity for a local adaptation process to reduce the unwanted outcomes after implementation. A workflow-oriented framework developed in a previous study indicates that users' needs and preferences could be incorporated into the system when implementation follows the steps of the framework, eventually increasing satisfaction with and usefulness of the system. The overall goal of this study was to demonstrate application of the workflow-oriented framework to the implementation of a nursing documentation system at Spaulding Rehabilitation Hospital. In this case study, we present specific steps of implementing and adapting a health information system at a local site and raise critical questions that need to be answered in each step based on the workflow-oriented framework.