Search

The negative impact of fire smoke on human health is widely recognized. However, studies have not adequately compared subtle differences in rates of emergency department (ED) visits and hospital admissions (HA) for various diseases, nor have they grounded comparisons in differences in healthcare systems across countries. To bridge this gap, we conducted a systematic meta-analysis to assess global health risks associated with fire smoke exposure. We performed an up-to-date systematic review by searching PubMed and Web of Science databases, using ED visits and HA as the main hospital indicators, and meta-analyzed the estimated risks. Our findings revealed an overall odds ratio (OR) of 1.07 (95 % CI: 1.06, 1.07) for global health risks (7 % increase) post-exposure to fire smoke. Specifically, the OR increased by 9 % for ED visits (OR = 1.09; 95 % CI: 1.08, 1.10) and by 4 % for HA (OR = 1.04; 95 % CI: 1.03, 1.05). When analyzing data by country, we noted that fire smoke exposure was associated with heightened disease risk in Sweden (OR = 1.63; 95 % CI: 1.22, 2.17), Canada (OR = 1.27; 95 % CI: 1.10, 1.46), the United States (OR = 1.10; 95 % CI: 1.08, 1.11), Brazil (OR = 1.07; 95 % CI: 1.04, 1.10), and Australia (OR = 1.04; 95 % CI: 1.03, 1.04). By continent, Europe exhibited the highest risk increase at 63 %, followed by North America at 10 %, South America at 7 %, and Australia at 4 %. By disease type, the most prevalent conditions were respiratory diseases, with a 7 % increase for total respiratory diseases and chronic obstructive pulmonary disease (COPD), 5 % for other pulmonary diseases, and 11 % for asthma. Cardiovascular diseases also showed significant increases: 5 % for general cardiovascular issues, 18 % for heart failure, and 5 % for ischemic diseases. Additional health concerns included headaches (26 %), itching (28 %), and skin allergies and rashes (34 %). We also explored the toxicological mechanisms behind these disease risks, the impact of climate variations, and the disparities in healthcare systems across different countries. This study provides stakeholders such as academics and physicians with a systematic understanding of the quantifiable health risks associated with fire smoke exposure. Additionally, it innovatively compares the differences between ED visits and HA, and between environmental and health systems in each country. The results of this study provide a foundation for subsequent relevant academic research and offer easy-to-understand information for policymakers, healthcare organizations, and the public. Ultimately, this will contribute to a better understanding and mitigation of the health risks associated with fire smoke.

BackgroundThe use of social media for disseminating health care information has become increasingly prevalent, making the expanding role of artificial intelligence (AI) and machine learning in this process both significant and inevitable. This development raises numerous ethical concerns. This study explored the ethical use of AI and machine learning in the context of health care information on social media platforms (SMPs). It critically examined these technologies from the perspectives of fairness, accountability, transparency, and ethics (FATE), emphasizing computational and methodological approaches that ensure their responsible application. ObjectiveThis study aims to identify, compare, and synthesize existing solutions that address the components of FATE in AI applications in health care on SMPs. Through an in-depth exploration of computational methods, approaches, and evaluation metrics used in various initiatives, we sought to elucidate the current state of the art and identify existing gaps. Furthermore, we assessed the strength of the evidence supporting each identified solution and discussed the implications of our findings for future research and practice. In doing so, we made a unique contribution to the field by highlighting areas that require further exploration and innovation. MethodsOur research methodology involved a comprehensive literature search across PubMed, Web of Science, and Google Scholar. We used strategic searches through specific filters to identify relevant research papers published since 2012 focusing on the intersection and union of different literature sets. The inclusion criteria were centered on studies that primarily addressed FATE in health care discussions on SMPs; those presenting empirical results; and those covering definitions, computational methods, approaches, and evaluation metrics. ResultsOur findings present a nuanced breakdown of the FATE principles, aligning them where applicable with the American Medical Informatics Association ethical guidelines. By dividing these principles into dedicated sections, we detailed specific computational methods and conceptual approaches tailored to enforcing FATE in AI-driven health care on SMPs. This segmentation facilitated a deeper understanding of the intricate relationship among the FATE principles and highlighted the practical challenges encountered in their application. It underscored the pioneering contributions of our study to the discourse on ethical AI in health care on SMPs, emphasizing the complex interplay and the limitations faced in implementing these principles effectively. ConclusionsDespite the existence of diverse approaches and metrics to address FATE issues in AI for health care on SMPs, challenges persist. The application of these approaches often intersects with additional ethical considerations, occasionally leading to conflicts. Our review highlights the lack of a unified, comprehensive solution for fully and effectively integrating FATE principles in this domain. This gap necessitates careful consideration of the ethical trade-offs involved in deploying existing methods and underscores the need for ongoing research.

BackgroundHomelessness represents a widespread social issue globally, yet the risk of neurodegenerative diseases (NDDs) associated with people experiencing homelessness (PEH) has not received sufficient attention. Therefore, this study aimed to explore the risk of NDDs among PEH and its variation across countries and regions through meta-analysis and systematic review.MethodsSearching from databases such as PubMed and Web of Science, relevant research articles on PEH and NDDs were identified. After multiple screening, eight articles were selected for meta-analysis. Statistical methods and models were used to evaluate the association between PEH and NDDs, stratified by disease type and country.ResultsWe found that PEH had a 51% higher risk of NDDs (OR = 1.51 (95% CI: 1.21, 1.89)) compared with those with stable housing. Specifically, PEH had a significantly higher risk of developing multiple sclerosis (OR = 4.64 (95% CI: 1.96, 10.98)). Alzheimer's disease and related dementias (ADRD) (OR = 1.93 (95% CI: 1.34, 2.77)), dementia (OR = 1.69 (95% CI: 1.26, 2.27)), and cognitive impairment (OR = 1.07 (95% CI: 0.98, 1.16)) were all at higher risk. Furthermore, country and regional differences were observed, with countries such as Iran (OR = 4.64 (95% CI: 1.96, 10.98)), the Netherlands (OR = 2.14 (95% CI: 1.23, 3.73)), the United States (OR = 1.66 (95% CI: 1.25, 2.22)), and Canada (OR = 1.06 (95% CI: 1.01, 1.10)) showing a higher risk of NDDs among the PEH.ConclusionsThe study emphasizes the significant NDD risks among PEH, providing novel perspectives on this issue and shedding light on national disparities influenced by variations in healthcare systems and social environments. This will be beneficial for academia and government to prioritize the health of PEH with NDDs, aiming to mitigate disease incidence and economic burdens while preserving social stability and upholding basic human rights.

Data annotation in NLP is a costly and time-consuming task, traditionally handled by human experts who require extensive training to enhance the task-related background knowledge. Besides, labeling social media texts is particularly challenging due to their brevity, informality, creativity, and varying human perceptions regarding the sociocultural context of the world. With the emergence of GPT models and their proficiency in various NLP tasks, this study aims to establish a performance baseline for GPT-4 as a social media text annotator. To achieve this, we employ our own dataset of tweets, expertly labeled for stance detection with full inter-rater agreement among three annotators. We experiment with three techniques: Zero-shot, Few-shot, and Zero-shot with Chain-of-Thoughts to create prompts for the labeling task. We utilize four training sets constructed with different label sets, including human labels, to fine-tune transformer-based large language models and various combinations of traditional machine learning models with embeddings for stance classification. Finally, all fine-tuned models undergo evaluation using a common testing set with human-generated labels. We use the results from models trained on human labels as the benchmark to assess GPT-4's potential as an annotator across the three prompting techniques. Based on the experimental findings, GPT-4 achieves comparable results through the Few-shot and Zero-shot Chain-of-Thoughts prompting methods. However, none of these labeling techniques surpass the top three models fine-tuned on human labels. Moreover, we introduce the Zero-shot Chain-of-Thoughts as an effective strategy for aspect-based social media text labeling, which performs better than the standard Zero-shot and yields results similar to the high-performing yet expensive Few-shot approach.

BackgroundA hallmark of unregulated drug markets is their unpredictability and constant evolution with newly introduced substances. People who use drugs and the public health workforce are often unaware of the appearance of new drugs on the unregulated market and their type, safe dosage, and potential adverse effects. This increases risks to people who use drugs, including the risk of unknown consumption and unintentional drug poisoning. Early warning systems (EWSs) can help monitor the landscape of emerging drugs in a given community by collecting and tracking up-to-date information and determining trends. However, there are currently few ways to systematically monitor the appearance and harms of new drugs on the unregulated market in Canada. ObjectiveThe goal of this work is to examine how artificial intelligence can assist in identifying patterns of drug-related risks and harms, by monitoring the social media activity of public health and law enforcement groups. This information is beneficial in the form of an EWS as it can be used to identify new and emerging drug trends in various communities. MethodsTo collect data for this study, 145 relevant Twitter accounts throughout Quebec (n=33), Ontario (n=78), and British Columbia (n=34) were manually identified. Tweets posted between August 23 and December 21, 2021, were collected via the application programming interface developed by Twitter for a total of 40,393 tweets. Next, subject matter experts (1) developed keyword filters that reduced the data set to 3746 tweets and (2) manually identified relevant tweets for monitoring and early warning efforts for a total of 464 tweets. Using this information, a zero-shot classifier was applied to tweets from step 1 with a set of keep (drug arrest, drug discovery, and drug report) and not-keep (drug addiction support, public safety report, and others) labels to see how accurately it could extract the tweets identified in step 2. ResultsWhen looking at the accuracy in identifying relevant posts, the system extracted a total of 584 tweets and had an overlap of 392 out of 477 (specificity of ~84.5%) with the subject matter experts. Conversely, the system identified a total of 3162 irrelevant tweets and had an overlap of 3090 (sensitivity of ~94.1%) with the subject matter experts. ConclusionsThis study demonstrates the benefits of using artificial intelligence to assist in finding relevant tweets for an EWS. The results showed that it can be quite accurate in filtering out irrelevant information, which greatly reduces the amount of manual work required. Although the accuracy in retaining relevant information was observed to be lower, an analysis showed that the label definitions can impact the results significantly and would therefore be suitable for future work to refine. Nonetheless, the performance is promising and demonstrates the usefulness of artificial intelligence in this domain.

Abstract Background Twitter is a popular social networking site where short messages or “tweets” of users have been used extensively for research purposes. However, not much research has been done in mining the medical professions, such as detecting the occupations of users from their biographical contents. Mining such professions can be used to build efficient recommender systems for cost-effective targeted advertisements. Moreover, it is highly important to develop effective methods to identify the occupation of users since conventional classification methods rely on features developed by human intelligence. Although, the result may be favorable for the classification problem. However, it is still extremely challenging for traditional classifiers to predict the medical occupations accurately since it involves predicting multiple occupations. Hence this study emphasizes predicting the medical occupational class of users through their public biographical (“Bio”) content. We have conducted our analysis by annotating the bio content of Twitter users. In this paper, we propose a method of combining word embedding with state-of-art neural network models that include: Long Short Term Memory (LSTM), Bidirectional LSTM, Gated Recurrent Unit, Bidirectional Encoder Representations from Transformers, and A lite BERT. Moreover, we have also observed that by composing the word embedding with the neural network models there is no need to construct any particular attribute or feature. By using word embedding, the bio contents are formatted as dense vectors which are fed as input into the neural network models as a sequence of vectors. Result Performance metrics that include accuracy, precision, recall, and F1-score have shown a significant difference between our method of combining word embedding with neural network models than with the traditional methods. The scores have proved that our proposed approach has outperformed the traditional machine learning techniques for detecting medical occupations among users. ALBERT has performed the best among the deep learning networks with an F1 score of 0.90. Conclusion In this study, we have presented a novel method of detecting the occupations of Twitter users engaged in the medical domain by merging word embedding with state-of-art neural networks. The outcomes of our approach have demonstrated that our method can further advance the process of analyzing corpora of social media without going through the trouble of developing computationally expensive features.

BackgroundOver the past years, homelessness has become a substantial issue around the globe. The largest social services organization in Thunder Bay, Ontario, Canada, has observed that a majority of the people experiencing homelessness in the city were from outside of the city or province. Thus, to improve programming and resource allocation for people experiencing homelessness in the city, including shelter use, it was important to investigate the trends associated with homelessness and migration. ObjectiveThis study aimed to address 3 research questions related to homelessness and migration in Thunder Bay: What factors predict whether a person who migrated to the city and is experiencing homelessness stays or leaves shelters? If an individual stays, how long are they likely to stay? What factors predict stay duration? MethodsWe collected the required data from 2 sources: a survey conducted with people experiencing homelessness at 3 homeless shelters in Thunder Bay and the database of a homeless information management system. The records of 110 migrants were used for the analysis. Two feature selection techniques were used to address the first and third research questions, and 8 machine learning models were used to address the second research question. In addition, data augmentation was performed to improve the size of the data set and to resolve the class imbalance problem. The area under the receiver operating characteristic curve value and cross-validation accuracy were used to measure the models’ performances while avoiding possible model overfitting. ResultsFactors predicting an individual’s stay duration included home or previous district, highest educational qualification, recent receipt of mental health support, migrating to visit family or friends, and finding employment upon arrival. For research question 2, among the classification models developed for predicting the stay duration of migrants, the random forest and gradient boosting tree models presented better results with area under the receiver operating characteristic curve values of 0.91 and 0.93, respectively. Finally, home district, band membership, status card, previous district, and recent support for drug and/or alcohol use were recognized as the factors predicting stay duration. ConclusionsApplying machine learning enables researchers to make predictions related to migrants’ homelessness and investigate how various factors become determinants of the predictions. We hope that the findings of this study will aid future policy making and resource allocation to better serve people experiencing homelessness. However, further improvements in the data set size and interpretation of the identified factors in decision-making are required.

Abstract Background In a sudden cardiac arrest, starting CPR and applying an AED immediately are the two highest resuscitation priorities. Many existing mobile applications have been developed to assist users in locating a nearby AED. However, these applications do not provide indoor navigation to the AED location. The time required to locate an AED inside a building due to a lack of indoor navigation systems will reduce the patient’s chance of survival. The existing indoor navigation solutions either require special hardware, a large dataset or a significant amount of initial work. These requirements make these systems not viable for implementation on a large-scale. Methods The proposed system collects Wi-Fi information from the existing devices and the path’s magnetic information using a smartphone to guide the user from a starting point to an AED. The information collected is processed using four techniques: turn detection method, Magnetic data pattern matching method, Wi-Fi fingerprinting method and Closest Wi-Fi location method to estimate user location. The user location estimations from all four techniques are further processed to determine the user’s location on the path, which is then used to guide the user to the AED location. Results The four techniques used in the proposed system Turn detection, Magnetic data pattern matching, Closest Wi-Fi location and Wi-Fi fingerprinting can individually achieve the accuracy of 80% with the error distance ± 9.4 m, ± 2.4 m, ± 4.6 m, and ± 4.6 m respectively. These four techniques, applied individually, may not always provide stable results. Combining these techniques results in a robust system with an overall accuracy of 80% with an error distance of ± 2.74 m. In comparison, the proposed system’s accuracy is higher than the existing systems that use Wi-Fi and magnetic data. Conclusion This research proposes a novel approach that requires no special hardware, large scale data or significant initial work to provide indoor navigation. The proposed system AEDNav can achieve an accuracy similar to the existing indoor navigation systems. Implementing this indoor navigation system could reduce the time to locate an AED and ultimately increase patient survival during sudden cardiac arrest.

BackgroundIn most cases, the abstracts of articles in the medical domain are publicly available. Although these are accessible by everyone, they are hard to comprehend for a wider audience due to the complex medical vocabulary. Thus, simplifying these complex abstracts is essential to make medical research accessible to the general public. ObjectiveThis study aims to develop a deep learning–based text simplification (TS) approach that converts complex medical text into a simpler version while maintaining the quality of the generated text. MethodsA TS approach using reinforcement learning and transformer–based language models was developed. Relevance reward, Flesch-Kincaid reward, and lexical simplicity reward were optimized to help simplify jargon-dense complex medical paragraphs to their simpler versions while retaining the quality of the text. The model was trained using 3568 complex-simple medical paragraphs and evaluated on 480 paragraphs via the help of automated metrics and human annotation. ResultsThe proposed method outperformed previous baselines on Flesch-Kincaid scores (11.84) and achieved comparable performance with other baselines when measured using ROUGE-1 (0.39), ROUGE-2 (0.11), and SARI scores (0.40). Manual evaluation showed that percentage agreement between human annotators was more than 70% when factors such as fluency, coherence, and adequacy were considered. ConclusionsA unique medical TS approach is successfully developed that leverages reinforcement learning and accurately simplifies complex medical paragraphs, thereby increasing their readability. The proposed TS approach can be applied to automatically generate simplified text for complex medical text data, which would enhance the accessibility of biomedical research to a wider audience.

This study assesses the online societal association of leaders and healthcare organizations from the top-10 COVID-19 resilient nations through public engagement, sentiment strength, and inclusivity and diversity strength. After analyzing 173,071 Tweets authored by the leaders and health organizations, our findings indicate that United Arab Emirate’s Prime Minister had the highest online societal association (normalized online societal association: 1.000) followed by the leaders of Canada and Turkey (normalized online societal association: 0.068 and 0.033, respectively); and among the healthcare organizations, the Public Health Agency of Canada was the most impactful (normalized online societal association: 1.000) followed by the healthcare agencies of Turkey and Spain (normalized online societal association: 0.632 and 0.094 respectively). In comparison to healthcare organizations, the leaders displayed a strong awareness of individual factors and generalized their Tweets to a broader audience. The findings also suggest that users prefer accessing social media platforms for information during health emergencies and that leaders and healthcare institutions should realize the potential to use them effectively.

The use of Twitter by healthcare organizations is an effective means of disseminating medical information to the public. However, the content of tweets can be influenced by various factors, such as health emergencies and medical breakthroughs. In this study, we conducted a discourse analysis to better understand how public and private healthcare organizations use Twitter and the factors that influence the content of their tweets. Data were collected from the Twitter accounts of five private pharmaceutical companies, two US and two Canadian public health agencies, and the World Health Organization from 1 January 2020, to 31 December 2022. The study applied topic modeling and association rule mining to identify text patterns that influence the content of tweets across different Twitter accounts. The findings revealed that building a reputation on Twitter goes beyond just evaluating the popularity of a tweet in the online sphere. Topic modeling, when applied synchronously with hashtag and tagging analysis can provide an increase in tweet popularity. Additionally, the study showed differences in language use and style across the Twitter accounts’ categories and discussed how the impact of popular association rules could translate to significantly more user engagement. Overall, the results of this study provide insights into natural language processing for health literacy and present a way for organizations to structure their future content to ensure maximum public engagement.

BackgroundSocial media platforms (SMPs) are frequently used by various pharmaceutical companies, public health agencies, and nongovernment organizations (NGOs) for communicating health concerns, new advancements, and potential outbreaks. Although the benefits of using them as a tool have been extensively discussed, the online activity of various health care organizations on SMPs during COVID-19 in terms of engagement and sentiment forecasting has not been thoroughly investigated. ObjectiveThe purpose of this research is to analyze the nature of information shared on Twitter, understand the public engagement generated on it, and forecast the sentiment score for various organizations. MethodsData were collected from the Twitter handles of 5 pharmaceutical companies, 10 US and Canadian public health agencies, and the World Health Organization (WHO) from January 1, 2017, to December 31, 2021. A total of 181,469 tweets were divided into 2 phases for the analysis, before COVID-19 and during COVID-19, based on the confirmation of the first COVID-19 community transmission case in North America on February 26, 2020. We conducted content analysis to generate health-related topics using natural language processing (NLP)-based topic-modeling techniques, analyzed public engagement on Twitter, and performed sentiment forecasting using 16 univariate moving-average and machine learning (ML) models to understand the correlation between public opinion and tweet contents. ResultsWe utilized the topics modeled from the tweets authored by the health care organizations chosen for our analysis using nonnegative matrix factorization (NMF): cumass=–3.6530 and –3.7944 before and during COVID-19, respectively. The topics were chronic diseases, health research, community health care, medical trials, COVID-19, vaccination, nutrition and well-being, and mental health. In terms of user impact, WHO (user impact=4171.24) had the highest impact overall, followed by public health agencies, the Centers for Disease Control and Prevention (CDC; user impact=2895.87), and the National Institutes of Health (NIH; user impact=891.06). Among pharmaceutical companies, Pfizer’s user impact was the highest at 97.79. Furthermore, for sentiment forecasting, autoregressive integrated moving average (ARIMA) and seasonal autoregressive integrated moving average with exogenous factors (SARIMAX) models performed best on the majority of the subsets of data (divided as per the health care organization and period), with the mean absolute error (MAE) between 0.027 and 0.084, the mean square error (MSE) between 0.001 and 0.011, and the root-mean-square error (RMSE) between 0.031 and 0.105. ConclusionsOur findings indicate that people engage more on topics such as COVID-19 than medical trials and customer experience. In addition, there are notable differences in the user engagement levels across organizations. Global organizations, such as WHO, show wide variations in engagement levels over time. The sentiment forecasting method discussed presents a way for organizations to structure their future content to ensure maximum user engagement.

Semantic textual similarity is one of the open research challenges in the field of Natural Language Processing. Extensive research has been carried out in this field and near-perfect results are achieved by recent transformer-based models in existing benchmark datasets like the STS dataset and the SICK dataset. In this paper, we study the sentences in these datasets and analyze the sensitivity of various word embeddings with respect to the complexity of the sentences. In this article, we build a complex sentence dataset comprising of 50 sentence pairs with associated semantic similarity values provided by 15 human annotators. Readability analysis is performed to highlight the increase in complexity of the sentences in the existing benchmark datasets and those in the proposed dataset. Further, we perform a comparative analysis of the performance of various word embeddings and language models on the existing benchmark datasets and the proposed dataset. The results show the increase in complexity of the sentences has a significant impact on the performance of the embedding models resulting in a 10-20% decrease in Pearson’s and Spearman’s correlation.

When modelling epidemics, the outputs and techniques used may be hard for the general public to understand. This can cause fear mongering and confusion on how to interpret the predictions provided by these models. This article proposes a solution for such a model that was created by a Canadian institute for COVID-19 in their region; namely, the NorthCOVID-19 model. In taking these ethical concerns into consideration, first the web interface of this model is analyzed to see how it may be difficult for a user without a strong mathematical background to understand how to use it. Second, a system is developed that takes this model’s outputs as an input and produces a video summarization with an auto-generated audio to address the complexity of the interface, while ensuring that the end user is able to understand the important information produced by this model. A survey conducted on this proposed output asked participants, on a scale of 1 to 5, whether they strongly disagreed (1) or strongly agreed (5) with statements regarding the output of the proposed method. The results showed that the audio in the output was helpful in understanding the results (80% responded with 4 or 5) and that it helped improve overallcomprehension of the model (85% responded with 4 or 5). For the analysis of the NorthCOVID-19 interface, a System Usability Scale (SUS) survey was performed where itreceived a scoring of 70.94 which is slightly above the average of 68.

Abstract Background When an Out-of-Hospital Cardiac Arrest (OHCA) incident is reported to emergency services, the 911 agent dispatches Emergency Medical Services to the location and activates responder network system (RNS), if the option is available. The RNS notifies all the registered users in the vicinity of the cardiac arrest patient by sending alerts to their mobile devices, which contains the location of the emergency. The main objective of this research is to find the best match between the user who could support the OHCA patient. Methods For performing matching among the user and the AEDs, we used Bipartite Matching and Integer Linear Programming. However, these approaches take a longer processing time; therefore, a new method Preprocessed Integer Linear Programming is proposed that solves the problem faster than the other two techniques. Results The average processing time for the experimentation data was 1850 s using Bipartite matching, 32 s using the Integer Linear Programming and 2 s when using the Preprocessed Integer Linear Programming method. The proposed algorithm performs matching among users and AEDs faster than the existing matching algorithm and thus allowing it to be used in the real world. Conclusion: This research proposes an efficient algorithm that will allow matching of users with AED in real-time during cardiac emergency. Implementation of this system can help in reducing the time to resuscitate the patient.

Abstract Background The collection and examination of social media has become a useful mechanism for studying the mental activity and behavior tendencies of users. Through the analysis of a collected set of Twitter data, a model will be developed for predicting positively referenced, drug-related tweets. From this, trends and correlations can be determined. Methods Social media data (tweets and attributes) were collected and processed using topic pertaining keywords, such as drug slang and use-conditions (methods of drug consumption). Potential candidates were preprocessed resulting in a dataset of 3,696,150 rows. The predictive classification power of multiple methods was compared including SVM, XGBoost, BERT and CNN-based classifiers. For the latter, a deep learning approach was implemented to screen and analyze the semantic meaning of the tweets. Results To test the predictive capability of the model, SVM and XGBoost were first employed. The results calculated from the models respectively displayed an accuracy of 59.33% and 54.90%, with AUC’s of 0.87 and 0.71. The values show a low predictive capability with little discrimination. Conversely, the CNN-based classifiers presented a significant improvement, between the two models tested. The first was trained with 2661 manually labeled samples, while the other included synthetically generated tweets culminating in 12,142 samples. The accuracy scores were 76.35% and 82.31%, with an AUC of 0.90 and 0.91. Using association rule mining in conjunction with the CNN-based classifier showed a high likelihood for keywords such as “smoke”, “cocaine”, and “marijuana” triggering a drug-positive classification. Conclusion Predictive analysis with a CNN is promising, whereas attribute-based models presented little predictive capability and were not suitable for analyzing text of data. This research found that the commonly mentioned drugs had a level of correspondence with frequently used illicit substances, proving the practical usefulness of this system. Lastly, the synthetically generated set provided increased accuracy scores and improves the predictive capability.

It is estimated that over 235,000 Canadians experience homelessness at some point each year. With the emergence of smart cities, it would be beneficial to leverage the processing power of deep learning to assist in the planning and testing of different policies to address this issue. When examining a population of homeless individuals, one can view them as being distributed, at any one point in time, among several possible states: for example, the street or an emergency shelter. Our work aims to provide a means of simulating across these states, including no longer homeless, over time. The probability that an individual will transition from one state to another is called a transition probability. Thus, by creating a matrix of transition probabilities between all of the states, we have a transition probability matrix. If we simply approached this problem by using a mathematical model such as a Markov decision process, we run into the issue of how to accurately adjust the probabilities to produce realistic results. Ideally, we would have a model that can reasonably modify them based on real-life data. To do this, we introduce two modified deep learning algorithms; modified deep q-learning (MDQL) and modified neural fitted q-iteration (MNFQ). These algorithms dynamically produce a set of transition probability matrices for each week of the year. We discuss the modifications we made to these algorithms to adapt to the homelessness problem and create our simulation. After training our model on high resolution, weekly data, we will show that when running it on a low resolution data set that spans 3 years, our model is able to achieve a relative percent difference from the final population of 12.5%. The end result is a model that can be further improved over time with real world data to provide realistic results.

The smart city is a concept of utilizing digital technologies to improve and enhance the lives of a city's inhabitants. This concept has been the subject of increasing interest over the past few years. However, most studies address improving aspects of a city's infrastructure, such as information security, privacy, communication networks, government, and transportation. Noticeably absent from the subject matter of these studies are social problems, such as poverty and homelessness. In this paper, we explore how technology can be harnessed to mitigate homelessness. We introduce eight novel heuristic algorithms that create a desirable homeless-to-housing assignment with regards to homeless individuals' characteristics and the nature of services. We discuss the efficiency of each of the algorithms through simulations. Our best performing algorithm obtains 92% accuracy in comparison to the optimal solution and 99.7% fairness. The algorithms are compared in terms of execution time, solution accuracy, fairness, and the relative difference with the optimal solution of this NP-hard problem.

Overcrowding is a well-known problem in hospitals and emergency departments (ED) that can negatively impact patients and staff. This study aims to present a machine learning model to detect a patient’s need for a Computed Tomography (CT) exam in the emergency department at the earliest possible time. The data for this work was collected from ED at Thunder Bay Regional Health Sciences Centre over one year (05/2016-05/2017) and contained administrative triage information. The target outcome was whether or not a patient required a CT exam. Multiple combinations of text embedding methods, machine learning algorithms, and data resampling methods were experimented with to find the optimal model for this task. The final model was trained with 81, 118 visits and tested on a hold-out test set with a size of 9, 013 visits. The best model achieved a ROC AUC score of 0.86 and had a sensitivity of 87.3% and specificity of 70.9%. The most important factors that led to a CT scan order were found to be chief complaint, treatment area, and triage acuity. The proposed model was able to successfully identify patients needing a CT using administrative triage data that is available at the initial stage of a patient’s arrival. By determining that a CT scan is needed early in the patient’s visit, the ED can allocate resources to ensure these investigations are completed quickly and patient flow is maintained to reduce overcrowding.

Emotion recognition in conversations is an important step in various virtual chatbots which require opinion-based feedback, like in social media threads, online support, and many more applications. Current emotion recognition in conversations models face issues like: (a) loss of contextual information in between two dialogues of a conversation, (b) failure to give appropriate importance to significant tokens in each utterance, (c) inability to pass on the emotional information from previous utterances. The proposed model of Advanced Contextual Feature Extraction (AdCOFE) addresses these issues by performing unique feature extraction using knowledge graphs, sentiment lexicons and phrases of natural language at all levels (word and position embedding) of the utterances. Experiments on emotion recognition in conversations datasets show that AdCOFE is beneficial in capturing emotions in conversations.

The semantic analysis field has a crucial role to play in the research related to text analytics. Calculating the semantic similarity between sentences is a long-standing problem in the area of natural language processing, and it differs significantly as the domain of operation differs. In this paper, we present a methodology that can be applied across multiple domains by incorporating corpora-based statistics into a standardized semantic similarity algorithm. To calculate the semantic similarity between words and sentences, the proposed method follows an edge-based approach using a lexical database. When tested on both benchmark standards and mean human similarity dataset, the methodology achieves a high correlation value for both word (r = 0.8753) and sentence similarity (r = 0.8793) concerning Rubenstein and Goodenough standard and the SICK dataset (r = 0.83241) outperforming other unsupervised models.

Within the field of post-secondary student mobility, the assessment, and evaluation of transfer credit is a labor-intensive human intelligence task that is subject to time limits and human bias. This paper introduces a semi-automated approach to assessing transfer credit and generating articulation agreements between post-secondary institutions using natural language processing (NLP). The output from the NLP system is tested using a content expert generated an assessment of transfer credit between computer science programs at two separate post-secondary institutions. Initial testing with an unsupervised NLP algorithm, despite good results against standardized measures, assessed the percentage of course overlap as 71% similar to the percentages selected by human content experts. The application of an algorithm based on the Word2Vec model using domain-specific Wikipedia corpus and dependency parsing was applied to compensate for domain specific language and improved the relationship between content experts ratings and NLP output to 86% related overlap.

Manual grading of essays by humans is time-consuming and likely to be susceptible to inconsistencies and inaccuracies. In recent years, an abundance of research has been done to automate essay evaluation processes, yet little has been done to take into consideration the syntax, semantic coherence and sentiments of the essay's text together. Our proposed system incorporates not just the rule-based grammar and surface level coherence check but also includes the semantic similarity of the sentences. We propose to use Graph-based relationships within the essay's content and polarity of opinion expressions. Semantic similarity is determined between each statement of the essay to form these Graph-based spatial relationships and novel features are obtained from it. Our algorithm uses 23 salient features with high predictive power, which is less than the current systems while considering every aspect to cover the dimensions that a human grader focuses on. Fewer features help us get rid of the redundancies of the data so that the predictions are based on more representative features and are robust to noisy data. The prediction of the scores is done with neural networks using the data released by the ASAP competition held by Kaggle. The resulting agreement between human grader's score and the system's prediction is measured using Quadratic Weighted Kappa (QWK). Our system produces a QWK of 0.793.

When conducting data analysis in the twenty-first century, social media is crucial to the analysis due to the ability to provide information on a variety of topics such as health, food, feedback on products, and many others. Presently, users utilize social media to share their daily lifestyles. For example, travel locations, exercises, and food are common subjects of social media posts. By analyzing such information collected from users, health of the general population can be gauged. This analysis can become an integral part of federal efforts to study the health of a nation's people on a large scale. In this paper, we focus on such efforts from a Canadian lens. Public health is becoming a primary concern for many governments around the world. It is believed that it is necessary to analyze the current scenario within a given population before creating any new policies. Traditionally, governments use a variety of ways to gauge the flavor for any new policy including door to door surveys, a national level census, or hospital information to decide health policies. This information is limited and sometimes takes a long time to collect and analyze sufficiently enough to aid in decision making. In this paper, our approach is to solve such problems through the advancement of natural language processing algorithms and large scale data analysis. Our in-depth results show that the proposed method provides a viable solution in less time with the same accuracy when compared to traditional methods.