Factors influencing presentation delay among cancer patients: a cross-sectional study in Malaysia.

BACKGROUND: Cancer represents a significant global public health challenge, with escalating incidence rates straining healthcare systems. Malaysia, like many nations, has witnessed a rise in cancer cases, particularly among the younger population. This study aligns with Malaysia's National Strategic Plan for Cancer Control Programme 2021-2025, emphasizing primary prevention and early detection to address cancer's impact. Therefore, we aim to describe the timeliness of cancer care for symptom presentation, socio-demographic, patient, as well as organizational-related factors among patients in Malaysia diagnosed with breast, colorectal, nasopharyngeal, and cervical cancer. METHODS: This cross-sectional study enrolled adult cancer patients diagnosed with breast, cervical, colorectal, or nasopharyngeal cancer from 2015 to 2020 in seven public hospitals/oncology centres across Malaysia. Data were collected through patient-administered surveys and medical records. Presentation delay, defined as the duration between symptom onset and the patient's first visit to a healthcare professional exceeding 30 days, was the primary outcome. Statistical analysis included descriptive statistics and chi-square tests. RESULTS: The study included 476 cancer patients, with breast cancer (41.6%), colorectal cancer (26.9%), nasopharyngeal cancer (22.1%), and cervical cancer (9.5%). Over half (54.2%) experienced presentation delays with a median interval of 60 days. Higher proportions of presentation delay were observed among nasopharyngeal cancer patients, employed patients with lower socioeconomic statuses, and those without family history of cancer. Most patients self-discovered their first cancer symptoms (80%), while only one-third took immediate action for medical check-ups. Emotional and organizational factors, such as long waiting times during doctor's visits (47%), were potential barriers to seeking cancer care. CONCLUSION: This study highlights the significant problem of presentation delay among cancer patients in Malaysia. The delay is influenced by various factors encompassing sociodemographic characteristics, health-seeking behaviours, and healthcare system-related issues. A comprehensive approach addressing both individual barriers and institutional obstacles is imperative to mitigate this presentation delay and improve cancer outcomes.

OpenDeID Pipeline for Unstructured Electronic Health Record Text Notes Based on Rules and Transformers: Deidentification Algorithm Development and Validation Study.

BACKGROUND: Electronic health records (EHRs) in unstructured formats are valuable sources of information for research in both the clinical and biomedical domains. However, before such records can be used for research purposes, sensitive health information (SHI) must be removed in several cases to protect patient privacy. Rule-based and machine learning-based methods have been shown to be effective in deidentification. However, very few studies investigated the combination of transformer-based language models and rules. OBJECTIVE: The objective of this study is to develop a hybrid deidentification pipeline for Australian EHR text notes using rules and transformers. The study also aims to investigate the impact of pretrained word embedding and transformer-based language models. METHODS: In this study, we present a hybrid deidentification pipeline called OpenDeID, which is developed using an Australian multicenter EHR-based corpus called OpenDeID Corpus. The OpenDeID corpus consists of 2100 pathology reports with 38,414 SHI entities from 1833 patients. The OpenDeID pipeline incorporates a hybrid approach of associative rules, supervised deep learning, and pretrained language models. RESULTS: The OpenDeID achieved a best F1-score of 0.9659 by fine-tuning the Discharge Summary BioBERT model and incorporating various preprocessing and postprocessing rules. The OpenDeID pipeline has been deployed at a large tertiary teaching hospital and has processed over 8000 unstructured EHR text notes in real time. CONCLUSIONS: The OpenDeID pipeline is a hybrid deidentification pipeline to deidentify SHI entities in unstructured EHR text notes. The pipeline has been evaluated on a large multicenter corpus. External validation will be undertaken as part of our future work to evaluate the effectiveness of the OpenDeID pipeline.

The need to strengthen the evaluation of the impact of Artificial Intelligence-based decision support systems on healthcare provision.

Despite the renewed interest in Artificial Intelligence-based clinical decision support systems (AI-CDS), there is still a lack of empirical evidence supporting their effectiveness. This underscores the need for rigorous and continuous evaluation and monitoring of processes and outcomes associated with the introduction of health information technology. We illustrate how the emergence of AI-CDS has helped to bring to the fore the critical importance of evaluation principles and action regarding all health information technology applications, as these hitherto have received limited attention. Key aspects include assessment of design, implementation and adoption contexts; ensuring systems support and optimise human performance (which in turn requires understanding clinical and system logics); and ensuring that design of systems prioritises ethics, equity, effectiveness, and outcomes. Going forward, information technology strategy, implementation and assessment need to actively incorporate these dimensions. International policy makers, regulators and strategic decision makers in implementing organisations therefore need to be cognisant of these aspects and incorporate them in decision-making and in prioritising investment. In particular, the emphasis needs to be on stronger and more evidence-based evaluation surrounding system limitations and risks as well as optimisation of outcomes, whilst ensuring learning and contextual review. Otherwise, there is a risk that applications will be sub-optimally embodied in health systems with unintended consequences and without yielding intended benefits.

The Role of Formative Evaluation in Promoting Digitally-based Health Equity and Reducing Bias for Resilient Health Systems: The Case of Patient Portals.

OBJECTIVES: Patient portals are increasingly implemented to improve patient involvement and engagement. We here seek to provide an overview of ways to mitigate existing concerns that these technologies increase inequity and bias and do not reach those who could benefit most from them. METHODS: Based on the current literature, we review the limitations of existing evaluations of patient portals in relation to addressing health equity, literacy and bias; outline challenges evaluators face when conducting such evaluations; and suggest methodological approaches that may address existing shortcomings. RESULTS: Various stakeholder needs should be addressed before deploying patient portals, involving vulnerable groups in user-centred design, and studying unanticipated consequences and impacts of information systems in use over time. CONCLUSIONS: Formative approaches to evaluation can help to address existing shortcomings and facilitate the development and implementation of patient portals in an equitable way thereby promoting the creation of resilient health systems.

Identifying and addressing digital health risks associated with emergency pandemic response: Problem identification, scoping review, and directions toward evidence-based evaluation.

BACKGROUND AND OBJECTIVE: The COVID-19 pandemic has accelerated digital health applications in multifaceted disease management dimensions. This study aims (1) to identify risk issues relating to the rapid development and redeployment of COVID-19 related e-health systems, in primary care, and in the health ecosystems interacting with it and (2) to suggest evidence-based evaluation directions under emergency response. METHOD: After initial brainstorming of digital health risks posed in this pandemic, a scoping review method was adopted to collect evidence across databases of PubMed, CINAHL, and EMBASE. Peer-review publications, reports, news sources, and websites that credibly identified the challenges relating digital health scaled for COVID-19 were scrutinized. Additional supporting materials were obtained through snowball sampling and the authors' global digital health networks. Studies satisfying the selection criteria were charted based on their study design, primary care focus, and coverage of e-health areas of risk. RESULTS: Fifty-eight studies were mapped for qualitative synthesis. Five identified digital health risk areas associated with the pandemic were governance, system design and coordination, information access, service provision, and user (professional and public) reception. We observed that rapid digital health responses may embed challenges in health system thinking, the long-term development of digital health ecosystems, and interoperability of health IT infrastructure, with concomitant weaknesses in existing evaluation theories. CONCLUSION: Through identifying digital health risks posed during the pandemic, this paper discussed potential directions for next-generation informatics evaluation development, to better prepare for the post-COVID-19 era, a new future epidemic, or other unforeseen global health emergencies. An updated evidence-based approach to health informatics is essential to gain public confidence in digital health across primary and other health sectors.

"Japan LIVE Dashboard" for COVID-19: A Scalable Solution to Monitor Real-Time and Regional-Level Epidemic Case Data.

Under pandemic conditions, it is important to communicate local infection risks to better enable the general population to adjust their behaviors accordingly. In Japan, our team operates a popular non-government and not-for-profit dashboard project - "Japan LIVE Dashboard" - which allows the public to easily grasp the evolution of the pandemic on the internet. We presented the Dashboard design concept with a generic framework integrating socio-technical theories, disease epidemiology and related contexts, and evidence-based approaches. Through synthesizing multiple types of reliable and real-time local data sources from all prefectures across the country, the Dashboard allows the public access to user-friendly and intuitive disease visualization in real time and has gained an extensive online followership. To date, it has attracted c.30 million visits (98% domestic access) testifying to the reputation it has acquired as a user-friendly portal for understanding the progression of the pandemic. Designed as an open-source solution, the Dashboard can also be adopted by other countries as well as made applicable for other emerging outbreaks in the future. Furthermore, the conceptual design framework may prove applicable into other ehealth scaled for global pandemics.

Managing Pandemic Responses with Health Informatics - Challenges for Assessing Digital Health Technologies.

OBJECTIVES: To highlight the role of technology assessment in the management of the COVID-19 pandemic. METHOD: An overview of existing research and evaluation approaches along with expert perspectives drawn from the International Medical Informatics Association (IMIA) Working Group on Technology Assessment and Quality Development in Health Informatics and the European Federation for Medical Informatics (EFMI) Working Group for Assessment of Health Information Systems. RESULTS: Evaluation of digital health technologies for COVID-19 should be based on their technical maturity as well as the scale of implementation. For mature technologies like telehealth whose efficacy has been previously demonstrated, pragmatic, rapid evaluation using the complex systems paradigm which accounts for multiple sociotechnical factors, might be more suitable to examine their effectiveness and emerging safety concerns in new settings. New technologies, particularly those intended for use on a large scale such as digital contract tracing, will require assessment of their usability as well as performance prior to deployment, after which evaluation should shift to using a complex systems paradigm to examine the value of information provided. The success of a digital health technology is dependent on the value of information it provides relative to the sociotechnical context of the setting where it is implemented. CONCLUSION: Commitment to evaluation using the evidence-based medicine and complex systems paradigms will be critical to ensuring safe and effective use of digital health technologies for COVID-19 and future pandemics. There is an inherent tension between evaluation and the imperative to urgently deploy solutions that needs to be negotiated.

Medication-rights detection using incident reports: A natural language processing and deep neural network approach.

Medication errors often occurred due to the breach of medication rights that are the right patient, the right drug, the right time, the right dose and the right route. The aim of this study was to develop a medication-rights detection system using natural language processing and deep neural networks to automate medication-incident identification using free-text incident reports. We assessed the performance of deep neural network models in classifying the Advanced Incident Reporting System reports and compared the models' performance with that of other common classification methods (including logistic regression, support vector machines and the decision-tree method). We also evaluated the effects on prediction outcomes of several deep neural network model settings, including number of layers, number of neurons and activation regularisation functions. The accuracy of the models was measured at 0.9 or above across model settings and algorithms. The average values obtained for accuracy and area under the curve were 0.940 (standard deviation: 0.011) and 0.911 (standard deviation: 0.019), respectively. It is shown that deep neural network models were more accurate than the other classifiers across all of the tested class labels (including wrong patient, wrong drug, wrong time, wrong dose and wrong route). The deep neural network method outperformed other binary classifiers and our default base case model, and parameter arguments setting generally performed well for the five medication-rights datasets. The medication-rights detection system developed in this study successfully uses a natural language processing and deep-learning approach to classify patient-safety incidents using the Advanced Incident Reporting System reports, which may be transferable to other mandatory and voluntary incident reporting systems worldwide.

Exploring Hidden In-Hospital Fall Clusters from Incident Reports Using Text Analytics.

Retrospective analysing of fall incident reports can uncover hidden information, identify potential risk factors, and improve healthcare quality. This study explores potential fall incident clusters using word embeddings and hierarchical clustering. Fall incident reports from 7 local hospitals in Hong Kong were catalogued into 5 potential clusters with significantly different fall severity, gender, reporting department, and keywords. This study demonstrates the feasibility of using text clustering methods on real-world fall incident reports mining.

Classification Scheme for Incident Reports of Medication Errors.

This study aimed to develop a classification scheme for retrieving information from incident reports of medication errors. This 15-category classification scheme captures minimal medication-incident related information from incident reports and thus serves as an information model for automatic information retrieval solution. The automatic solution uses recent advances in artificial intelligence methods to learn from incident report resources and is promising to the prevention of adverse drug events and promotion of safety in medical care.

Artificial Intelligence in Clinical Decision Support: Challenges for Evaluating AI and Practical Implications.

OBJECTIVES: This paper draws attention to: i) key considerations for evaluating artificial intelligence (AI) enabled clinical decision support; and ii) challenges and practical implications of AI design, development, selection, use, and ongoing surveillance. METHOD: A narrative review of existing research and evaluation approaches along with expert perspectives drawn from the International Medical Informatics Association (IMIA) Working Group on Technology Assessment and Quality Development in Health Informatics and the European Federation for Medical Informatics (EFMI) Working Group for Assessment of Health Information Systems. RESULTS: There is a rich history and tradition of evaluating AI in healthcare. While evaluators can learn from past efforts, and build on best practice evaluation frameworks and methodologies, questions remain about how to evaluate the safety and effectiveness of AI that dynamically harness vast amounts of genomic, biomarker, phenotype, electronic record, and care delivery data from across health systems. This paper first provides a historical perspective about the evaluation of AI in healthcare. It then examines key challenges of evaluating AI-enabled clinical decision support during design, development, selection, use, and ongoing surveillance. Practical aspects of evaluating AI in healthcare, including approaches to evaluation and indicators to monitor AI are also discussed. CONCLUSION: Commitment to rigorous initial and ongoing evaluation will be critical to ensuring the safe and effective integration of AI in complex sociotechnical settings. Specific enhancements that are required for the new generation of AI-enabled clinical decision support will emerge through practical application.

Long-term Results of Mini Asymmetric Radial Keratotomy and Corneal Cross-linking for the Treatment of Keratoconus.

PURPOSE: To investigate the long-term results (at least 5 years of follow-up) of the mini asymmetric radial keratotomy (MARK) and corneal cross-linking (CXL) combined intervention, also known as the 'Rome protocol,' for patients with progressive stage I and II keratoconus and contact lens intolerance. METHODS: This was a retrospective observational case series. Fifteen eyes of 12 patients were evaluated, with a mean follow-up of 6.9 years. To assess the efficacy and stability of the MARK + CXL combined protocol, best spectacle-corrected visual acuity, mean pachymetry, and mean keratometry were recorded preoperatively and at least 1, 3, and 5 years postoperatively. Statistical analysis was performed using the R platform and involved the Wilcoxon signed-rank and Kruskal-Wallis non-parametric tests. RESULTS: Best spectacle-corrected visual acuity improved for all patients, from 0.46 ± 0.69 logarithm of the minimum angle of resolution (20 / 60) to 0.15 ± 0.69 logarithm of the minimum angle of resolution (20 / 30, p = 0.0006), while mean pachymetry increased in 93% of patients, from 442.80 ± 61.02 to 464.50 ± 62.72 µm (p = 0.003). Lastly, mean keratometry improved in 87% of patients after 6.9 years of observation from 48.82 ± 5.00 to 43.25 ± 3.58 diopters (p = 0.008). No intraoperative or postoperative complications were observed. CONCLUSIONS: The MARK + CXL combined protocol was effective in treating keratoconus by halting corneal thinning and bulging. In addition, this procedure significantly improved visual acuity based on long-term follow-up data. Analysis of data from a larger cohort of patients would be useful to support these findings.

A Framework of Rebalancing Imbalanced Healthcare Data for Rare Events' Classification: A Case of Look-Alike Sound-Alike Mix-Up Incident Detection.

Identifying rare but significant healthcare events in massive unstructured datasets has become a common task in healthcare data analytics. However, imbalanced class distribution in many practical datasets greatly hampers the detection of rare events, as most classification methods implicitly assume an equal occurrence of classes and are designed to maximize the overall classification accuracy. In this study, we develop a framework for learning healthcare data with imbalanced distribution via incorporating different rebalancing strategies. The evaluation results showed that the developed framework can significantly improve the detection accuracy of medical incidents due to look-alike sound-alike (LASA) mix-ups. Specifically, logistic regression combined with the synthetic minority oversampling technique (SMOTE) produces the best detection results, with a significant 45.3% increase in recall (recall = 75.7%) compared with pure logistic regression (recall = 52.1%).

Temporal prediction of in-hospital falls using tensor factorisation.

In-hospital fall incidence is a critical indicator of healthcare outcome. Predictive models for fall incidents could facilitate optimal resource planning and allocation for healthcare providers. In this paper, we proposed a tensor factorisation-based framework to capture the latent features for fall incidents prediction over time. Experiments with real-world data from local hospitals in Hong Kong demonstrated that the proposed method could predict the fall incidents reasonably well (with an area under the curve score around 0.9). As compared with the baseline time series models, the proposed tensor based models were able to successfully identify high-risk locations without records of fall incidents during the past few months.

The Safe and Effective Use of Shared Data Underpinned by Stakeholder Engagement and Evaluation Practice.

OBJECTIVES: The paper draws attention to: i) key considerations involving the confidentiality, privacy, and security of shared data; and ii) the requirements needed to build collaborative arrangements encompassing all stakeholders with the goal of ensuring safe, secure, and quality use of shared data. METHOD: A narrative review of existing research and policy approaches along with expert perspectives drawn from the International Medical Informatics Association (IMIA) Working Group on Technology Assessment and Quality Development in Health Care and the European Federation for Medical Informatics (EFMI) Working Group for Assessment of Health Information Systems. RESULTS: The technological ability to merge, link, re-use, and exchange data has outpaced the establishment of policies, procedures, and processes to monitor the ethics and legality of shared use of data. Questions remain about how to guarantee the security of shared data, and how to establish and maintain public trust across large-scale shared data enterprises. This paper identifies the importance of data governance frameworks (incorporating engagement with all stakeholders) to underpin the management of the ethics and legality of shared data use. The paper also provides some key considerations for the establishment of national approaches and measures to monitor compliance with best practice. CONCLUSION: Data sharing endeavours can help to underpin new collaborative models of health care which provide shared information, engagement, and accountability amongst all stakeholders. We believe that commitment to rigorous evaluation and stakeholder engagement will be critical to delivering health data benefits and the establishment of collaborative models of health care into the future.

Context Sensitive Health Informatics: Delivering 21st Century Healthcare - Building a Quality-and-Efficiency Driven System.

Health informatics applications will be a cornerstone in the next generation quality-and-efficiency health care system. Health care is delivered from many different specialties, to many different patients with complex diseases and comorbidity. A one size fits all approach is not adequate to reach the Triple Aim of improving the patient experience of care, improving the health of populations, and reducing the per capita cost of health care. Health informatics applications must be built to be adaptable and sensitive to the complex contexts they will be used in. To enhance patient-centeredness in the 21st Century healthcare, research attention should be focused on investigating and designing models contributing to effective health information retrieval process.

Uncovering Hidden Topics in Hong Kong Clinical Research Through Hospital Authority Convention Publications.

Uncovering clinical research trends allows us to understand the direction of healthcare services and is essential for longer-term healthcare planning. The Hospital Authority Convention is a mainstream annual healthcare conference that gathers up-to-date Hong Kong medical research. We propose to use state-of-the-art medical document mining and topic modelling methods to uncover latent themes and structures in the publications. We collected 742 articles from HA Convention from the year 2013 to 2016 and selected 56 publications from the category of "Clinical Safety and Quality Service" for further analysis. Applying natural language processing and Latent Dirichlet Allocation (LDA) methods, we identified 7 potential topics, namely: surgical operation, hospital discharge, medical error, nursing procedure, service performance assessment, patient and staff engagement, and admission algorithm and standardisation. This exploratory study demonstrates that key themes exist in the annual HA Convention and we observe potential changes in healthcare services focus over the years in the selected category.

Economic Evaluation of Individual School Closure Strategies: The Hong Kong 2009 H1N1 Pandemic.

BACKGROUND: School closures as a means of containing the spread of disease have received considerable attention from the public health community. Although they have been implemented during previous pandemics, the epidemiological and economic effects of the closure of individual schools remain unclear. METHODOLOGY: This study used data from the 2009 H1N1 pandemic in Hong Kong to develop a simulation model of an influenza pandemic with a localised population structure to provide scientific justifications for and economic evaluations of individual-level school closure strategies. FINDINGS: The estimated cost of the study's baseline scenario was USD330 million. We found that the individual school closure strategies that involved all types of schools and those that used a lower threshold to trigger school closures had the best performance. The best scenario resulted in an 80% decrease in the number of cases (i.e., prevention of about 830,000 cases), and the cost per case prevented by this intervention was USD1,145; thus, the total cost was USD1.28 billion. CONCLUSION: This study predicts the effects of individual school closure strategies on the 2009 H1N1 pandemic in Hong Kong. Further research could determine optimal strategies that combine various system-wide and district-wide school closures with individual school triggers across types of schools. The effects of different closure triggers at different phases of a pandemic should also be examined.

Statistical classification of drug incidents due to look-alike sound-alike mix-ups.

It has been recognised that medication names that look or sound similar are a cause of medication errors. This study builds statistical classifiers for identifying medication incidents due to look-alike sound-alike mix-ups. A total of 227 patient safety incident advisories related to medication were obtained from the Canadian Patient Safety Institute's Global Patient Safety Alerts system. Eight feature selection strategies based on frequent terms, frequent drug terms and constituent terms were performed. Statistical text classifiers based on logistic regression, support vector machines with linear, polynomial, radial-basis and sigmoid kernels and decision tree were trained and tested. The models developed achieved an average accuracy of above 0.8 across all the model settings. The receiver operating characteristic curves indicated the classifiers performed reasonably well. The results obtained in this study suggest that statistical text classification can be a feasible method for identifying medication incidents due to look-alike sound-alike mix-ups based on a database of advisories from Global Patient Safety Alerts.

Statistical text classifier to detect specific type of medical incidents.

WHO Patient Safety has put focus to increase the coherence and expressiveness of patient safety classification with the foundation of International Classification for Patient Safety (ICPS). Text classification and statistical approaches has showed to be successful to identifysafety problems in the Aviation industryusing incident text information. It has been challenging to comprehend the taxonomy of medical incidents in a structured manner. Independent reporting mechanisms for patient safety incidents have been established in the UK, Canada, Australia, Japan, Hong Kong etc. This research demonstrates the potential to construct statistical text classifiers to detect specific type of medical incidents using incident text data. An illustrative example for classifying look-alike sound-alike (LASA) medication incidents using structured text from 227 advisories related to medication errors from Global Patient Safety Alerts (GPSA) is shown in this poster presentation. The classifier was built using logistic regression model. ROC curve and the AUC value indicated that this is a satisfactory good model.