Cognitive Health IT 2020 Abstracts


Area 1 - Cognitive Health IT

Full Papers
Paper Nr: 2
Title:

Automated Rheumatic Heart Disease Detection from Phonocardiogram in Cardiology Ward

Authors:

Melkamu H. Asmare, Frehiwot Woldehanna, Luc Janssens and Bart Vanrumste

Abstract: Rheumatic Heart Disease (RHD) is a preventable and treatable form of cardiovascular diseases. It is also referred to as the ailment of the disadvantaged mainly affecting children and young adults. RHD is recognized as a global health priority by World Health Organization. This chronic heart condition silently deteriorates the normal function of the heart valves which can be detected as a heart murmur using a stethoscope. As the cardiac auscultation process is an elusive process, the clinician will always be tempted to refer the patient for expensive and sophisticated imaging procedures like echocardiography. In this study, a machine learning algorithm is developed to augment the limitation in the auscultation process and transform the stethoscope as a powerful screening tool. For this current study, an RHD heart sound data set is recorded from one hundred seventy subjects. A total of twenty-six features are extracted to model murmur due to RHD. Twenty-four classification and regression algorithms have been tested out of which the Cubic SVM has demonstrated superiority with a classification accuracy of 97.1%, with 98% sensitivity, 95.3 % of specificity 97.6% precision. The corresponding positive predictive values (PPV) are 96% and 97% for normal and RHD respectively. The results are based on data collected from a cardiology ward where there are more pathological cases than controls. Hence it is a valuable detection tool in a cardiology clinic. But in the future, integrating this machine learning algorithm with a mobile phone can be a powerful screening tool in places where access to echocardiography and cardiologist is difficult. Thus, it can then aid a timely, affordable and reliable detection tool allowing a non-medically trained individual to screen and detect RHD.

Paper Nr: 4
Title:

AI-Rehab: A Framework for AI Driven Neurorehabilitation Training - The Profiling Challenge

Authors:

Yuri Almeida, Manisha S. Sirsat, Sergi Bermúdez i Badia and Eduardo Fermé

Abstract: One of the health clinic challenges is rehabilitation therapy cognitive impairment that can happen after brain injury, dementia and in normal cognitive decline due to aging. Current cognitive rehabilitation therapy has been shown to be the most effective way to address this problem. However, a) it is not adaptive for every patient, b) it has a high cost, and c) it is usually implemented in clinical environments. The Task Generator (TG) is a free tool for the generation of cognitive training tasks. However, TG is not designed to adapt and monitor the cognitive progress of the patient. Hence, we propose in the BRaNT project an enhancement of TG with belief revision and machine learning techniques, gamification and remote monitoring capabilities to enable health professionals to provide a long-term personalized cognitive rehabilitation therapy at home. The BRaNT is an interdisciplinary effort that addresses scientific limitations of current practices as well as provides solutions towards the sustainability of health systems and contributes towards the improvement of quality of life of patients. This paper proposes the AI-Rehab framework for the BRaNT, explains profiling challenge in the situation of insufficient data and presents an alternate AI solutions which might be applicable once enough data is available.

Paper Nr: 6
Title:

Using BERT and Semantic Patterns to Analyze Disease Outbreak Context over Social Network Data

Authors:

Neelesh Rastogi and Fazel Keshtkar

Abstract: Predicting disease outbreaks has been a focus for various institutions and researchers for many years. However, any models that seemed to get close to resolve this issue have failed to predict potential outbreaks with accuracy over time. For leveraging the social media data effectively, it is crucial to filter out noisy information from the large volume of data flux so that we could better estimate potential disease outbreaks with growing social data. Not satisfied with essential keyword-based filtration, many researchers turn to machine learning for a solution. In this paper, we apply deep learning techniques to address the Tweets classification problem concerning disease outbreak predictions. To achieve this, we curated a labeled corpus of Tweets that reflect different types of disease-related reports, showcasing diverse community sentiment and varied potential usages in emergency responses. Further, we used BERT, a word embedding and deep learning method to apply transfer learning against our curated dataset. Applying BERT showed that it performs better in comparable results to Long short-term memory (LSTM) and outperforming the baseline model on average in terms of accuracy and F-score.

Paper Nr: 7
Title:

Predicting 30-days All-cause Hospital Readmissions Considering Discharge-to-alternate-care-facilities

Authors:

Tahir Hameed and Syed C. Bukhari

Abstract: Hospital discharge is a decision based on several data points including diagnostic, physiological, demographic and caretaker information. Readmissions days after discharge are costly in addition to negative impact on capacity and service quality of hospitals. 30-days readmission (30DRA) literature remains focused on above variables and medical conditions paying little attention to the role of alternate-care-facilities (such as skilled nursing facilities and hospices) on reduction of 30DRA rates. To the best of our knowledge, there is negligible research considering alternate care variables for predicting readmissions even when physicians have actively started considering discharge-to-alternate-care during discharge planning. This paper develops a classification model for predicting patients who are likely to be readmitted within 30 days of discharge-to-alternate-care. Several machine-learning approaches, such as multi-logistic regression, Naïve Bayes, random forest, and neural networks were tested on the model to find the one with highest predictive power. The model was trained and tested on MIMIC-III, a large anonymized electronic health records (EHRs) database from US hospitals. Results suggest discharge-to-alternate-care reduces 30DRA. Moreover, neural networks and logistic regression techniques show better precision and accuracy in identifying the patients likely to be readmitted in 30 days.