Explore the vast range of titles available.

Background Topic models are a class of unsupervised machine learning models, which facilitate summarization, browsing and retrieval from large unstructured document collections. This study reviews several methods for assessing the quality of unsupervised topic models estimated using non-negative matrix factorization. Techniques for topic model validation have been developed across disparate fields. We synthesize this literature, discuss the advantages and disadvantages of different techniques for topic model validation, and illustrate their usefulness for guiding model selection on a large clinical text corpus. Design, setting and data Using a retrospective cohort design, we curated a text corpus containing 382,666 clinical notes collected between 01/01/2017 through 12/31/2020 from primary care electronic medical records in Toronto Canada. Methods Several topic model quality metrics have been proposed to assess different aspects of model fit. We explored the following metrics: reconstruction error, topic coherence, rank biased overlap, Kendall’s weighted tau, partition coefficient, partition entropy and the Xie-Beni statistic. Depending on context, cross-validation and/or bootstrap stability analysis were used to estimate these metrics on our corpus. Results Cross-validated reconstruction error favored large topic models (K ≥ 100 topics) on our corpus. Stability analysis using topic coherence and the Xie-Beni statistic also favored large models (K = 100 topics). Rank biased overlap and Kendall’s weighted tau favored small models (K = 5 topics). Few model evaluation metrics suggested mid-sized topic models (25 ≤ K ≤ 75) as being optimal. However, human judgement suggested that mid-sized topic models produced expressive low-dimensional summarizations of the corpus. Conclusions Topic model quality indices are transparent quantitative tools for guiding model selection and evaluation. Our empirical illustration demonstrated that different topic model quality indices favor models of different complexity; and may not select models aligning with human judgment. This suggests that different metrics capture different aspects of model goodness of fit. A combination of topic model quality indices, coupled with human validation, may be useful in appraising unsupervised topic models.

Health care organizations are collecting increasing volumes of clinical text data. Topic models are a class of unsupervised machine learning algorithms for discovering latent thematic patterns in these large unstructured document collections. We aimed to comparatively evaluate several methods for estimating temporal topic models using clinical notes obtained from primary care electronic medical records from Ontario, Canada. We used a retrospective closed cohort design. The study spanned from January 01, 2011, through December 31, 2015, discretized into 20 quarterly periods. Patients were included in the study if they generated at least 1 primary care clinical note in each of the 20 quarterly periods. These patients represented a unique cohort of individuals engaging in high-frequency use of the primary care system. The following temporal topic modeling algorithms were fitted to the clinical note corpus: nonnegative matrix factorization, latent Dirichlet allocation, the structural topic model, and the BERTopic model. Temporal topic models consistently identified latent topical patterns in the clinical note corpus. The learned topical bases identified meaningful activities conducted by the primary health care system. Latent topics displaying near-constant temporal dynamics were consistently estimated across models (eg, pain, hypertension, diabetes, sleep, mood, anxiety, and depression). Several topics displayed predictable seasonal patterns over the study period (eg, respiratory disease and influenza immunization programs). Nonnegative matrix factorization, latent Dirichlet allocation, structural topic model, and BERTopic are based on different underlying statistical frameworks (eg, linear algebra and optimization, Bayesian graphical models, and neural embeddings), require tuning unique hyperparameters (optimizers, priors, etc), and have distinct computational requirements (data structures, computational hardware, etc). Despite the heterogeneity in statistical methodology, the learned latent topical summarizations and their temporal evolution over the study period were consistently estimated. Temporal topic models represent an interesting class of models for characterizing and monitoring the primary health care system.

Background Due to the complexity of tumor genetic heterogeneity, personalized medicine has progressively emerged as the central focus of cancer research. However, how to accurately predict the drug response of patients before receiving treatment is the critical challenge to the development of this field. Methods This paper proposes DrugBERT, a BERT-based framework integrated with LDA topic embedding and a drug efficacy-aware mechanism for predicting the efficacy of antitumor drugs. The method incorporates LDA-generated topic embedding as a semantic enhancement module into the BERT language model and introduces a drug efficacy-aware attention mechanism to prioritize drug efficacy-related semantic features. The model is via LSTM to capture long-range dependencies in clinical text data. In addition, the SMOTE algorithm is used to synthesize samples of the minority class to solve the problem of data imbalance. Results The proposed method DrugBERT demonstrated remarkable performance on a dataset of 958 patients with non-small cell cancer treated with antitumor drugs. Furthermore, when validated on an independent dataset of 266 bowel cancer patients, the model achieved a 3% improvement in AUC over previous methods, signifying its robust generalization capability. Conclusions DrugBERT can help predict the efficacy of antitumor drugs based on clinical text while exhibiting strong generalization capability. These findings highlight its potential for optimizing personalized therapeutic strategies through language model.

Mobile technology has the potential to deliver behavior change interventions (mHealth) to reduce coronary heart disease (CHD) at modest cost. Previous studies have focused on single behaviors; however, cardiac rehabilitation (CR), a component of CHD self-management, needs to address multiple risk factors. The aim was to investigate the effectiveness of a mHealth-delivered comprehensive CR program (Text4Heart) to improve adherence to recommended lifestyle behaviors (smoking cessation, physical activity, healthy diet, and nonharmful alcohol use) in addition to usual care (traditional CR). A 2-arm, parallel, randomized controlled trial was conducted in New Zealand adults diagnosed with CHD. Participants were recruited in-hospital and were encouraged to attend center-based CR (usual care control). In addition, the intervention group received a personalized 24-week mHealth program, framed in social cognitive theory, sent by fully automated daily short message service (SMS) text messages and a supporting website. The primary outcome was adherence to healthy lifestyle behaviors measured using a self-reported composite health behavior score (≥3) at 3 and 6 months. Secondary outcomes included clinical outcomes, medication adherence score, self-efficacy, illness perceptions, and anxiety and/or depression at 6 months. Baseline and 6-month follow-up assessments (unblinded) were conducted in person. Eligible patients (N=123) recruited from 2 large metropolitan hospitals were randomized to the intervention (n=61) or the control (n=62) group. Participants were predominantly male (100/123, 81.3%), New Zealand European (73/123, 59.3%), with a mean age of 59.5 (SD 11.1) years. A significant treatment effect in favor of the intervention was observed for the primary outcome at 3 months (AOR 2.55, 95% CI 1.12-5.84; P=.03), but not at 6 months (AOR 1.93, 95% CI 0.83-4.53; P=.13). The intervention group reported significantly greater medication adherence score (mean difference: 0.58, 95% CI 0.19-0.97; P=.004). The majority of intervention participants reported reading all their text messages (52/61, 85%). The number of visits to the website per person ranged from zero to 100 (median 3) over the 6-month intervention period. A mHealth CR intervention plus usual care showed a positive effect on adherence to multiple lifestyle behavior changes at 3 months in New Zealand adults with CHD compared to usual care alone. The effect was not sustained to the end of the 6-month intervention. A larger study is needed to determine the size of the effect in the longer term and whether the change in behavior reduces adverse cardiovascular events. ACTRN 12613000901707; https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=364758&isReview=true (Archived by WebCite at http://www.webcitation.org/6c4qhcHKt).

The rapid uptake of mobile phones in low and middle-income countries over the past decade has provided public health programs unprecedented access to patients. While programs have used text messages to improve medication adherence, there have been no high-powered trials evaluating their impact on tuberculosis treatment outcomes. To measure the impact of Zindagi SMS, a two-way SMS reminder system, on treatment success of people with drug-sensitive tuberculosis. We conducted a two-arm, parallel design, effectiveness randomized controlled trial in Karachi, Pakistan. Individual participants were randomized to either Zindagi SMS or the control group. Zindagi SMS sent daily SMS reminders to participants and asked them to respond through SMS or missed (unbilled) calls after taking their medication. Non-respondents were sent up to three reminders a day. Public and private sector tuberculosis clinics in Karachi, Pakistan. Newly-diagnosed patients with smear or bacteriologically positive pulmonary tuberculosis who were on treatment for less than two weeks; 15 years of age or older; reported having access to a mobile phone; and intended to live in Karachi throughout treatment were eligible to participate. We enrolled 2,207 participants, with 1,110 randomized to Zindagi SMS and 1,097 to the control group. The primary outcome was clinically recorded treatment success based upon intention-to-treat. We found no significant difference between the Zindagi SMS or control groups for treatment success (719 or 83% vs. 903 or 83%, respectively, p = 0·782). There was no significant program effect on self-reported medication adherence reported during unannounced visits during treatment. In this large-scale randomized controlled effectiveness trial of SMS medication reminders for tuberculosis treatment, we found no significant impact. The trial was registered with ClinicalTrials.gov, NCT01690754.

Mobile phone text messages (short message service, SMS) are used pervasively as a form of communication. Almost 100% of the population uses text messaging worldwide and this technology is being suggested as a promising tool in psychiatry. Text messages can be sent either from a classic mobile phone or a web-based application. Reviews are needed to better understand how text messaging can be used in mental health care and other fields of medicine. The objective of the study was to review the literature regarding the use of mobile phone text messaging in mental health care. We conducted a thorough literature review of studies involving text messaging in health care management. Searches included PubMed, PsycINFO, Cochrane, Scopus, Embase and Web of Science databases on May 25, 2015. Studies reporting the use of text messaging as a tool in managing patients with mental health disorders were included. Given the heterogeneity of studies, this review was summarized using a descriptive approach. From 677 initial citations, 36 studies were included in the review. Text messaging was used in a wide range of mental health situations, notably substance abuse (31%), schizophrenia (22%), and affective disorders (17%). We identified four ways in which text messages were used: reminders (14%), information (17%), supportive messages (42%), and self-monitoring procedures (42%). Applications were sometimes combined. We report growing interest in text messaging since 2006. Text messages have been proposed as a health care tool in a wide spectrum of psychiatric disorders including substance abuse, schizophrenia, affective disorders, and suicide prevention. Most papers described pilot studies, while some randomized clinical trials (RCTs) were also reported. Overall, a positive attitude toward text messages was reported. RCTs reported improved treatment adherence and symptom surveillance. Other positive points included an increase in appointment attendance and in satisfaction with management and health care services. Insight into message content, preventative strategies, and innovative approaches derived from the mental health field may be applicable in other medical specialties.

Depressive disorders have substantial global implications, leading to various social consequences, including decreased occupational productivity and a high disability burden. Early detection and intervention for clinically significant depression have gained attention; however, the existing depression screening tools, such as the Center for Epidemiologic Studies Depression Scale, have limitations in objectivity and accuracy. Therefore, researchers are identifying objective indicators of depression, including image analysis, blood biomarkers, and ecological momentary assessments (EMAs). Among EMAs, user-generated text data, particularly from diary writing, have emerged as a clinically significant and analyzable source for detecting or diagnosing depression, leveraging advancements in large language models such as ChatGPT. We aimed to detect depression based on user-generated diary text through an emotional diary writing app using a large language model (LLM). We aimed to validate the value of the semistructured diary text data as an EMA data source. Participants were assessed for depression using the Patient Health Questionnaire and suicide risk was evaluated using the Beck Scale for Suicide Ideation before starting and after completing the 2-week diary writing period. The text data from the daily diaries were also used in the analysis. The performance of leading LLMs, such as ChatGPT with GPT-3.5 and GPT-4, was assessed with and without GPT-3.5 fine-tuning on the training data set. The model performance comparison involved the use of chain-of-thought and zero-shot prompting to analyze the text structure and content. We used 428 diaries from 91 participants; GPT-3.5 fine-tuning demonstrated superior performance in depression detection, achieving an accuracy of 0.902 and a specificity of 0.955. However, the balanced accuracy was the highest (0.844) for GPT-3.5 without fine-tuning and prompt techniques; it displayed a recall of 0.929. Both GPT-3.5 and GPT-4.0 demonstrated relatively reasonable performance in recognizing the risk of depression based on diaries. Our findings highlight the potential clinical usefulness of user-generated text data for detecting depression. In addition to measurable indicators, such as step count and physical activity, future research should increasingly emphasize qualitative digital expression.

Electronic health records (EHRs) consist of both structured data (eg, diagnostic codes) and unstructured data (eg, clinical notes). It is commonly believed that unstructured clinical narratives provide more comprehensive information. However, this assumption lacks large-scale validation and direct validation methods. This study aims to quantitatively compare the information in structured and unstructured EHR data and directly validate whether unstructured data offers more extensive information across a patient population. We analyzed both structured and unstructured data from patient records and visits in a large Dutch primary care EHR database between January 2021 and January 2024. Clinical concepts were identified from free-text notes using an extraction framework tailored for Dutch and compared with concepts from structured data. Concept embeddings were generated to measure semantic similarity between structured and extracted concepts through cosine similarity. A similarity threshold was systematically determined via annotated matches and minimized weighted Gini impurity. We then quantified the concept overlap between structured and unstructured data across various concept domains and patient populations. In a population of 1.8 million patients, only 13% of extracted concepts from patient records and 7% from individual visits had similar structured counterparts. Conversely, 42% of structured concepts in records and 25% in visits had similar matches in unstructured data. Condition concepts had the highest overlap, followed by measurements and drug concepts. Subpopulation visits, such as those with chronic conditions or psychological disorders, showed different proportions of data overlap, indicating varied reliance on structured versus unstructured data across clinical contexts. Our study demonstrates the feasibility of quantifying the information difference between structured and unstructured data, showing that the unstructured data provides important additional information in the studied database and populations. The annotated concept matches are made publicly available for the clinical natural language processing community. Despite some limitations, our proposed methodology proves versatile, and its application can lead to more robust and insightful observational clinical research.

Mobile phone SMS text message reminders have shown moderate effects in improving participation rates in ongoing colorectal cancer screening programs. This study aimed to assess the effectiveness of SMS text messages as a replacement for routine postal reminders in a fecal immunochemical test-based colorectal cancer screening program in Catalonia, Spain. We conducted a randomized controlled trial among individuals aged 50 to 69 years who were invited to screening but had not completed their fecal immunochemical test within 6 weeks. The intervention group (n=12,167) received an SMS text message reminder, while the control group (n=12,221) followed the standard procedure of receiving a reminder letter. The primary outcome was participation within 18 weeks of the invitation. The trial was stopped early, and a recovery strategy was implemented for nonparticipants in the intervention group. We performed a final analysis to evaluate the impact of the recovery strategy on the main outcome of the trial. Participation was assessed using a logistic regression model adjusting for potential confounders (sex, age, and deprivation score index) globally and by screening behavior. The trial was discontinued early in September 2022 due to the results of the interim analysis. The interim analysis included 5570 individuals who had completed 18 weeks of follow-up (intention-to-treat). The SMS text message group had a participation rate of 17.2% (477/2781), whereas the control group had a participation rate of 21.9% (610/2789; odds ratio 0.71, 95% CI 0.62-0.82; P<.001). As a recovery strategy, 7591 (72.7%) out of 10,442 nonparticipants in the SMS text message group had an open screening episode and received a second reminder by letter, reaching a participation rate of 23% (1748/7591). The final analysis (N=24,388) showed a participation rate of 29.3% (3561/12,167) in the intervention group, which received 2 reminders, while the participation rate was 26.5% (3235/12,221) in the control group (odds ratio 1.16, 95% CI 1.09-1.23; P<.001). Replacing SMS text messages with reminder letters did not increase the participation rate but also led to a decline in participation among nonparticipants 6 weeks after the invitation. However, sending a second reminder by letter significantly increased participation rates among nonparticipants within 6 weeks in the SMS text message group compared with those who received 1 postal reminder (control group). Additional research is essential to determine the best timing and frequency of reminders to boost participation without being intrusive in their choice of participation. ClinicalTrials.gov NCT04343950; https://www.clinicaltrials.gov/study/NCT04343950.

The initial introduction of the World Wide Web in 1990 brought around the biggest change in information acquisition. Due to the abundance of devices and ease of access they subsequently allow, the utility of mobile health (mHealth) has never been more endemic. A substantial amount of interactive and psychoeducational apps are readily available to download concerning a wide range of health issues. mHealth has the potential to reduce waiting times for appointments; eradicate the need to meet in person with a clinician, successively diminishing the workload of mental health professionals; be more cost effective to practices; and encourage self-care tactics. Previous research has given valid evidence with empirical studies proving the effectiveness of physical and mental health interventions using mobile apps. Alongside apps, there is evidence to show that receiving short message service (SMS) messages, which entail psychoeducation, medication reminders, and links to useful informative Web pages can also be advantageous to a patient's mental and physical well-being. Available mHealth apps and SMS services and their ever improving quality necessitates a systematic review in the area in reference to reduction of symptomology, adherence to intervention, and usability. The aim of this review was to study the efficacy, usability, and feasibility of mobile apps and SMS messages as mHealth interventions for self-guided care. A systematic literature search was carried out in JMIR, PubMed, PsychINFO, PsychARTICLES, Google Scholar, MEDLINE, and SAGE. The search spanned from January 2008 to January 2017. The primary outcome measures consisted of weight management, (pregnancy) smoking cessation, medication adherence, depression, anxiety and stress. Where possible, adherence, feasibility, and usability outcomes of the apps or SMS services were evaluated. Between-group and within-group effect sizes (Cohen d) for the mHealth intervention method group were determined. A total of 27 studies, inclusive of 4658 participants were reviewed. The papers included randomized controlled trials (RCTs) (n=19), within-group studies (n=7), and 1 within-group study with qualitative aspect. Studies show improvement in physical health and significant reductions of anxiety, stress, and depression. Within-group and between-group effect sizes ranged from 0.05-3.37 (immediately posttest), 0.05-3.25 (1-month follow-up), 0.08-3.08 (2-month follow-up), 0.00-3.10 (3-month follow-up), and 0.02-0.27 (6-month follow-up). Usability and feasibility of mHealth interventions, where reported, also gave promising, significant results. The review shows the promising and emerging efficacy of using mobile apps and SMS text messaging as mHealth interventions.