B8018 - DIGITAL HEALTH M

Learning outcomes

At the end of this interdisciplinary course unit, the student will acquire a deep knowledge on today's healthcare delivery systems (national and international ones) and their associated payer models, as well as on key determinants of health/wellness and the opportunities that digital health could bring based on these determinants. In addition, the student will be able to understand the primary models in the field, as well as the most relevant digital health technologies (from sensing and monitoring to medical devices, including software as a medical device, to mobile health apps, to health data science and health AI). By delving into finer details, the student will learn about how to re-think the digital health technology space into the directions of 'personal health informatics' and 'personal health cybernetics'. In addition, she will learn an advanced digital health technology/study in a specific disease/condition area (selected from a list provided by the instructor and also by the student via negotiation with the instructor). Finally, the student will learn through practice in designing and implementing an interdisciplinary group project, which will focus on either a personal health system or a health data science project. Lists of projects will be provided by the instructor, but student groups are welcome to propose their own projects, pending the instruction's approval.

Course contents

Healthcare and aging care are under economical and operational capacity pressures and are currently undergoing digital transformations, mainly utilizing several technologies, data science, AI, and digitalization to significantly cut unit and total cost of care, improve health outcomes, eradicate certain costly diseases, and in general, focus more on preventive and proactive measures (active and healthy living and aging through a continuum-of-care), than maintaining the status quo of a reactive disease management system (a point-of-care system).

This ambitious transformation is collectively referred to as “Digital Health” backed by an emerging and growing “Health Tech” industry. This emerging industry and the future health care delivery systems are needy of a skilled workforce equipped with the interdisciplinary knowledge and expertise to drive the implementation of such ambitious transformational changes. This interdisciplinary course on Digital Health entails aims to prepare such workforce of the future – the agents of change and the Digital Health leaders of the future.

Course Lectures:

1. Introduction to Digital Health (4 hours) Defining digital health from several perspectives. Providing different overviews addressing the different contexts of use and applicability of digital health including the broad array of digital health technologies. This module also provides an outline of all subsequent modules below.

2. Determinants of Health & Wellbeing (6 hours) What are the key determinants of a person's health, and which determinants stand to benefit the most from digital health? And how do we measure health and wellbeing? If we use digital health as a prevention or intervention, we should have a way to measure its effect - the health outcomes.

3. Healthcare Delivery Systems (6 hours) This module will shed light on the different care systems and their payer models in several parts of the world, spanning national and private care systems showing their unit and total cost and their pros and cons. This knowledge provides an essential context to the understanding of the roles digital health may play in anticipated future health transformations.

4. The Future of Health (6 hours) So how can digital health shape a better future for health? What would such a future look like in terms of health systems, focus, cost, center of gravity, technology, and services? We go through a "science fiction prototyping" (a visioning exercise) of such a future, and then zoom in to research-evident clues of the future of health as provided by specialized reports such as Deloitte Health Transformation group among others.

5. Group Research Project (4 hours) Systematic literature review of a digital health research/development/technology area of focus. This module will discuss the assignment and will provide a good example of a review completed by students in a prior offering of this course. Additionally, a tutorial (2 hours) on methodologies to conduct systematic research reviews will be given along with a video recording. 

6. Case Studies in Digital health (6 hours). We examine the current effect and influence of digital health through two case studies, one related to COPD (a pulmonary disease) and another to Atrial Fibrillation (a Cardiovascular disease).

7. Clinical Efficiency and Healthcare Delivery Improvements (4 hours) Covering how digital health can help improve the processes of care delivery (e.g., a consultation visits in the doctor's office, or performing a medical or diagnostic procedure in a hospital). Also, how to reduce unit and total cost and time taken in these processes. Clinical efficiency also addresses reductions of errors and personal bias in medical practice.

 

8. The Ethics and Regulatory Requirements for Conducting Digital Health Systems based Clinical Studies (6 hours): We briefly learn about the ethical issues relevant to research or product/prototype development that involves the participation of humans for any reason. We will focus on the required regulatory pathways for developing medical devices or software as a medical device (SaMD). You will specifically learn about the University of Bologna Ethics Review Committee (Comitato etico) process and regulatory policies including participants recruitment, consent, and ethics protocol design, privacy, and secure data management.

 

9. Health AI - Part I. (6 hours): Here you will recall basics of statistics particularly emphasizing descriptive and referential statistics as well as regression analysis. Then you will learn the basics of Machine Learning techniques including unsupervised learning (pattern discovery and clustering), and supervised learning (detection, classification, and prediction). Health datasets will be provided as well as tutorials (in addition to the lectures) to prepare you to analyze the data and develop models and train them to answer key classification or prediction questions. Python will be required but this is the easy part as you will only need to know/learn the very basics of Python in case you do not have prior Python experience (a tutorial for Python beginners will be provided (2 hours). More involved would be the Data Science libraries such as numpy, pandas, seaborn and matplotlib that you will need to learn and understand. Jupyter Notebook will also be required. You will be required to use Google Colab to avoid any individual installation issues on your laptop of Python 3.1, its libraries, Jupyter Notebook, and also to speed up your work particularly training and cross validation. A tutorial (3 hours) will be delivered off class hours to go over a full modeling exercise utilizing a fetal health dataset. You will enjoy this intense module if you are eager to learn and are motivated to be knowledgeable about Health Analytics and Health AI

 

10. Health AI - Part II. In this module, you will continue to learn about Health AI focusing on the basics of Deep Learning techniques suited for medical imaging data (classification an regression). Deep learning is a special type of machine learning which is particularly capable of handling large amounts of very complex data. You will particularly learn Convolutional Neural Networks (CNN) and get introduced to ImageNet before being introduced to Transfer Learning and Data Augmentation. Finally, you will get to see the limitation of CNN and introduced to the wonderful concept of Attention (instead of Convolution). You will learn the Transformer model in Vision (Vision Transformer, or ViT). A tutorial focusing on CNN and ViT along with their performance comparison will be delivered (4 hours). 

11. Group Hands-on Projects (4 hours) Several projects will be defined and offered in three thrust areas: mobile Health, Health AI and Health IoT (Internet of Things). Students will also be allowed to propose their own projects, but such proposals will have to be approved by Professor Helal and may be changed in scope. All projects will be discussed in the lectures of this module.

 

12. Health IoT, Wearables, and Smart Homes (9 hours) This module focuses on technologies including personal health systems where wearables, Health IoT (Internet of Things) and sensors are used to create systems for a variety of purposes such as monitoring, diagnosis, intervention, support and empowerment, compliance, among other goals. A particular kind of digital health system is Mobile Health in which mobile health apps are designed with similar goals in mind and the smartphone is used as a key component of such personal health systems. Mobile apps go beyond physical markers and vital signs that can be sensed by a pervasive system to include behavioral marker and markers related to lifestyles. Then we move to a more complex digital health systems which are health platforms such as smart homes and smart hospitals. We will focus on smart homes this term. The important difference in requirement and approach between devices, solutions, and platforms will be explained. Case studies will be provided.

13. Human Activity Recognition (HAR) (3 hours) This module focuses - briefly - on methods, models, and techniques to "making sense of sensor data" sourced from the users and their devices or platforms. We will cover several approaches to activity recognition including AI based, probabilistic, and the emerging LLM methods.   

14. Human Persuasive Cybernetics (3 hours). The Cybernetics of digital health systems represents the other direction of traffic where intervention, user/patient engagement, and health maintenance can be delivered and implemented. Ensuring user engagement and convergence is not easy and numerous “behavioral change models” have been proposed and studied. We will cover some of these models to the extent of clearly understanding the problem but not fully covering all available solution alternatives. 

Readings/Bibliography

In addition to class lecture slides, the instructor will provide a number of research and practice papers, technology reports, and datasets.

Teaching methods

The class will be taught through a series of lectures that are in-person (online and recording will be offered occasionally upon requests for students who have a good excuse for not being able to attend in person - but only occasionally).  Additionally, supplementary tutorials will be provided. In class activities, quizzes, and practice exams will be given. 

Assessment methods

Assessment will be based on individual interviews. All interviews will be held after the lecturing period ends. Students will be asked three questions each individually about the subjects studied to verify they meet the course learning objectives (about 15 min for each student).

Teaching tools

Microsoft Teams, Google Colab, lecture recordings, and Virtuale.

Office hours

See the website of Abdelsalam Ali Helal