Revision as of 10:02, 5 June 2025

Modules

AI In Healthcare: Technology Basics

Instructions for:TraineeTrainer

Goal

The aim

To support research ethics reviewers in learning about AI technologies for the review of projects and proposals that develop and/or use AI for healthcare.

The content focuses on key technology basics in a succinct manner, and signposts further learning opportunities for those who require more in-depth knowledge.

Learning outcomes

At the end of this module, learners will be able to:

Identify AI systems and how they are built.
Discuss some key applications of AI-based systems in healthcare.
Discuss the primary implications of the use of AI in healthcare.

Duration (hours)

2

For whom is this important?

All stakeholders in research

Part of

iRECS

Steps

1Opening Poll

2Antonija Mijatovic - Challenges for Data Privacy and Security

3AI and the Protection of Health Data - Activity

4AI and the Patient-Doctor Relationship

5New Actors in the Health Domain

12How are AI Systems Built?

13Key Applications of AI in the Healthcare Domain

14Alexei Grinbaum - the Benefits of AI

15Alexei Grinbaum - Reasons to be Cautious about the Use of AI

16Key Applications of AI In the Healthcare Domain Cont.

17AI and the Protection of Health Data

1

Opening Poll

Before you begin to work your way through the rest of this module, take a moment to think about how you feel about the increasing use of AI technologies in healthcare.

Are you excited and looking forward to seeing how it develops?
Are you against the use of AI technologies in healthcare?
Or do you have some reservations, but think it can be beneficial?

We will ask you to reflect on this again at the end of the module.

2

What is AI?

AI, or artificial intelligence, refers to the development of digital systems that can perform tasks that typically require human intelligence.

These tasks include learning, reasoning, problem-solving, perception, language understanding, and speech recognition.

At its core, AI leverages principles from computer science, mathematics, and cognitive psychology to replicate intelligent behaviour in machines.

AI utilises algorithms, data, and computational power to simulate intelligent behaviour, enabling machines to adapt, improve, and perform complex functions autonomously.

Several core scientific concepts underpin the development and functionality of AI. Work your way through the presentation below to hear about some of them:

This list of core scientific concepts in AI is subject to ongoing research and development. The field of AI is rapidly evolving, and new techniques, algorithms, and applications are continuously emerging.

As researchers and scientists make advancements in AI technology and explore novel use cases, the understanding and implementation of these concepts may evolve.

Generative AI is a type of foundation model that is becoming more and more evident in everyday life as well as in healthcare. Test your understanding of generative AI by answering the following questions.

AI In Healthcare: Technology Basics Step2_Audio Transcript2

3

What is AI? Quiz

What is the primary function of generative AI?

AI In Healthcare: Technology Basics Step3_quiz

4

How are AI Systems Built?

AI systems are built through a process that involves several key steps.

Building AI systems is an iterative process, that involves ongoing refinement and improvement to keep up with evolving requirements and challenges.

Additionally, advancements in research and technology may prompt updates to models and algorithms.

AI In Healthcare: Technology Basics Step4_Audio Transcript

5

Key Applications of AI in the Healthcare Domain

Artificial Intelligence (AI) is revolutionising the healthcare domain, introducing transformative applications that can enhance diagnosis, treatment, and patient care. The synergy between advanced algorithms, machine learning, and healthcare data has paved the way for innovative solutions with the potential to improve outcomes, increase efficiency, and streamline processes.

Key applications of AI in the healthcare sector include:

Diagnostic imagery: AI excels in interpreting medical imaging data, such as X-rays, MRIs, and CT scans. Deep learning algorithms can detect patterns and abnormalities, aiding radiologists in accurate and swift diagnoses. This not only speeds up the diagnostic process but also enhances precision when identifying subtle anomalies.

Disease prediction & prevention: Predictive analytics powered by AI enables healthcare providers to forecast disease risks and identify individuals who may be predisposed to certain conditions. By analysing patient data and lifestyle factors, AI models can assist in implementing preventive measures and personalised interventions to mitigate potential health risks.

Drug discovery & development: AI can accelerate the drug discovery process by analysing vast datasets to identify potential drug candidates and predict their efficacy. Machine learning algorithms can analyse molecular structures, predict drug interactions, and optimise formulations, significantly reducing the time and cost associated with bringing new drugs to market.

Personalised medicine: AI enables the development of personalised treatment plans by analysing patient-specific data, including genetic information, medical history, and lifestyle factors. This approach allows healthcare providers to tailor interventions and medications to individual patient needs, improving treatment effectiveness and minimising adverse effects.

Chatbots: Chatbots and virtual health assistants powered by AI are used to enhance patient engagement and provide on-demand healthcare information. These tools can offer guidance on symptoms, medication reminders, and lifestyle recommendations. Proponents of virtual health assistants claim that they can improve patient adherence to treatment plans and foster better communication between patients and healthcare providers. However, this is disputed by those who are concerned about what may be lost through a decrease in human-to-human interactions.

Natural language processing and health records: The use of NLP algorithms to extract valuable insights from unstructured clinical notes and electronic health records facilitates efficient data management, enables faster information retrieval, and supports clinical decision-making. NLP also plays a crucial role in automating administrative tasks, allowing healthcare professionals to focus more on patient care.

Robotic surgery assistance: AI-powered robotic systems assist surgeons in performing complex procedures with precision and minimal invasiveness. These robotic platforms enhance surgical outcomes, reduce recovery times, and contribute to advancements in minimally invasive surgery techniques.

Remote patient monitoring: AI can be used to facilitate the continuous monitoring of patients' health remotely through wearable devices and sensors. Real-time data analysis allows healthcare providers to track vital signs, detect abnormalities, and intervene promptly, especially for patients with chronic conditions.

AI In Healthcare: Technology Basics Step5_Presentation Transcript

6

Alexei Grinbaum - the Benefits of AI

Alexei Grinbaum shares his thoughts on the benefits of AI.

Benefits of AI

Alexei Grinbaum – French Alternative Energies and Atomic Energy Commission (CEA)

AI in healthcare can bring two kinds of benefits. One for the patients, AI does non-human calculation. It can probably, through computation, discover things that we haven't discovered humanly, meaning new drugs, new treatments. And that is happening already. We are discovering new molecules that we had never thought about humanly. So that's for the patients.

Now for the doctors, for medical doctors, AI can probably free up a lot of time from doing routine things that the doctors don't enjoy doing, like writing reports or things like that AI can have a lot of benefits for all sorts of people involved in healthcare on different sides, for medical professionals and for the patients, but very different kinds of benefits.

AI In Healthcare: Technology Basics Step6_Video

7

Alexei Grinbaum - Reasons to be Cautious about the Use of AI

Alexei Grinbaum shares his thoughts on reasons to be cautious about AI.

Reasons to be cautious about the use of AI

Like with all artificial intelligence systems, there should be limits, filters, and controls. We shouldn't just let it go uncontrolled completely. So unchecked completely. First, of course, there is the classic question of data, personal data, sensitive data. This is about health, our data. Some of it is genetic. Some of it lets us identify people. How do we treat that data? So, that is a very classic question. But beyond that, there are very interesting questions about human autonomy. AI systems, will they overtake the doctors? Will they still leave a place for human contact, human warmth?

If we seek advice from an AI system, does it mean that somehow the medical profession is changing completely? So, these kinds of questions are important. Again, they're not exactly specific to the medical sphere. They also exist, for example, for AI assistance. But in the medical sphere, there are interesting questions that are a little bit more specific. Some would say, cybersecurity, you know, it's everywhere. We have all heard about robustness and cybersecurity. But in the medical sphere, if you have a device that is interacting with your body, and if somebody can hack it, well then of course, it's a direct threat to our well-being, to our health. So, the questions of cybersecurity are also very touchy, I would say, in the medical sphere.

And then beyond that, we have classic big questions about organisations. Will the whole sphere of medical care with the hospitals, you know, the emergency rooms and all of these things, how will that evolve? Should it be managed not by humans, but by robots or AI systems? Will they respond faster? How will that change the way we build our social institutions? And that's another dimension of AI in health care.

So, there are definitely benefits at each of these levels. But there are also risks, or I would say, reasons to be cautious, reasons not to go too fast with the deployment of AI systems, because the human profession doesn't change in 10 days, right? We need time to evolve. We need time to learn new skills.

So not going too fast, teaching medical doctors and healthcare professionals to work together with these systems rather than be replaced by these systems. That is something that is very important. Take time. Take the time. Be cautious about bias, discrimination, accessibility, autonomy, control, all of those different things, and not go too fast.

AI In Healthcare: Technology Basics Step7_Video

8

Key Applications of AI In the Healthcare Domain Cont.

What are some of the key benefits of the use of AI applications in healthcare?

As AI continues to evolve, its applications in healthcare hold immense promise for improving the quality of care, optimising resource utilization, and potentially shaping a more patient-centered and efficient healthcare ecosystem. However, careful implementation and ongoing scrutiny is vital to ensure the responsible and beneficial use of AI in healthcare.

If you would like to explore the topic of AI implementation in more depth, you can access the Coalition for Health’s Blueprint for Trustworthy AI Implementation Guidance and Assurance for Healthcare in the further resources section.

AI In Healthcare: Technology Basics Step8_Audio Transcript

9

AI and the Protection of Health Data

As healthcare organizations strive to harness the benefits of AI while safeguarding patient data, a delicate balance must be maintained.

Stringent security protocols, transparent data governance, and ethical AI development practices are crucial for upholding the confidentiality and integrity of healthcare data in the era of AI-driven advancements.

Policymakers, healthcare providers, and technology developers must work collaboratively to establish robust frameworks, regulations, and ethical guidelines to ensure the responsible and secure use of health data in AI applications.

AI In Healthcare: Technology Basics Step9_Video

10

Antonija Mijatovic - Challenges for Data Privacy and Security

Antonija Mijatovic shares her thoughts on challenges for data privacy and security.

Challenges for data privacy and security

When it comes to data security and privacy, the major issues are data breaches. Because many applications in AI involve health data, and health data is sensitive and confidential by nature.

So, data breaches can lead to privacy violations, identity theft, even health risks. And they result in financial losses for healthcare organizations. Because healthcare is the top industry targeted by ransomware. Ransomware is a common cyber-attack. But aside of ransomware, data breaches can occur through hacking, phishing, and even if a device storing health information is lost or stolen. And data breaches can also happen unintentionally. For example, if patient data is emailed to the wrong recipient or posted online. And these incidents happen very often.

For example, in the United States alone, only in the last year there have been more than 500 cases of cyber-attacks. So, this is why it is important to address. Researchers need to take multiple measures to ensure data security and privacy. And these include cyber security measures, such as strong passwords, restricted access, two-factor authentication, and even encryption of very sensitive data. In addition, researchers should create backups of very important folders. And also, because 90% of cyber-attacks were allowed due to human error, researchers who work with sensitive data should receive proper training in the subject.

Ethics reviewers need to check whether researchers took all necessary measures to ensure data privacy and security. And they should also check whether researchers adhered to regulatory compliance. For example, in the European Union, personal data is regulated through the GDPR and personal data in AI is regulated through the Artificial Intelligence Act. While in the United States there are several guidelines such as the Health Accountability and Portability Act.

AI In Healthcare: Technology Basics Step10_Video about Expert Interview

11

AI and the Protection of Health Data - Activity

Think about places where data relating to your own health might be stored – tick all that apply

12

AI and the Patient-Doctor Relationship

Consider the following questions to reflect on the potential impact of AI technologies in healthcare on the patient-doctor relationship:

Awareness and Trust:

Do you think patients may have concerns or reservations about relying on AI-driven insights over traditional doctor-patient interactions?

Communication and Understanding:

In what ways do you think AI technologies could enhance or hinder communication between patients and healthcare professionals?

Personalisation and Empathy:

Do you think the integration of AI could impact the empathetic aspects of the patient-doctor relationship? If so, in what ways?

Role of the Healthcare Professional:

What role do you envision for healthcare professionals in a future where AI technologies play a significant role in diagnosis and treatment?
How can doctors maintain their essential role as caregivers and decision-makers while working alongside AI systems?

Balancing Technology and Human Touch:

Reflect on the importance of finding a balance between AI technologies and the human touch in healthcare. How can technology enhance, rather than replace, the human connection between patients and healthcare providers?

13

New Actors in the Health Domain

In navigating the entry of new actors into the health domain, policymakers, healthcare providers, and technology companies must work together to ensure that innovations align with patient needs, adhere to ethical standards, and contribute to the overall improvement of healthcare delivery. Balancing innovation with regulatory oversight and patient protection remains a key challenge in this evolving landscape.

14

End of Module Poll

In this module we have considered key concepts associated with the use of AI technologies in healthcare, including how AI systems are built, some of the key applications available for use in healthcare, and the primary implications of their use within the healthcare domain.

Now you can try the end of module quiz to see whether your learning from this module addresses the intended learning outcomes.

15

End of Module Quiz

You can try these questions to see whether your learning from this module addresses the intended learning outcomes. No one else will see your answers. No personal data is collected.

16

Module Evaluation

Thank you for taking this irecs module!

Your feedback is very valuable to us and will help us to improve future training materials.

We would like to ask for your opinions:

1. To improve the irecs e-learning modules

2. For research purposes to evaluate the outcomes of the irecs project

To this end we have developed a short questionnaire, which will take from 5 to 10 minutes to answer.

Your anonymity is guaranteed; you won’t be asked to share identifying information or any sensitive information. Data will be handled and stored securely and will only be used for the purposes detailed above. You can find the questionnaire by clicking on the link below.

This link will take you to a new page; https://forms.office.com/e/cimWP1L4tx

Thank you!

17

Bibliography

Alzubaidi, Laith, Jinglan Zhang, Amjad J. Humaidi, Ayad Al-Dujaili, Ye Duan, Omran Al-Shamma, J. Santamaría, Mohammed A. Fadhel, Muthana Al-Amidie, et Laith Farhan. Review of deep learning: concepts, CNN architectures, challenges, applications, future directions. Journal of Big Data 8, no 1 (31 mars 2021): 53. https://doi.org/10.1186/s40537-021-00444-8.

Bertuzzi, L., ‘AI Act: EU Countries Headed to Tiered Approach on Foundation Models amid Broader Compromise’, Www.Euractiv.Com, October 17, 2023. https://www.euractiv.com/section/artificial-intelligence/news/ai-act-eu-countries-headed-to-tiered-approach-on-foundation-models-amid-broader-compromise/

Bommasani, R. et al. (2022) ‘On the Opportunities and Risks of Foundation Models’. arXiv. Available at: https://doi.org/10.48550/arXiv.2108.07258.

Castellino, Ronald A. « Computer aided detection (CAD): an overview. Cancer Imaging 5, no 1 (23 août 2005): 1719. https://doi.org/10.1102/1470-7330.2005.0018.

Chan, Heang-Ping, Lubomir M. Hadjiiski, et Ravi K. Samala. Computer-Aided Diagnosis in the Era of Deep Learning. Medical Physics 47, no 5 (juin 2020): e218-27. https://doi.org/10.1002/mp.13764.

Chapman, Benjamin P., Feng Lin, Shumita Roy, Ralph H. B. Benedict, et Jeffrey M. Lyness. Health risk prediction models incorporating personality data: Motivation, challenges, and illustration. Personality Disorders: Theory, Research, and Treatment 10, no 1 (2019): 46-58. https://doi.org/10.1037/per0000300.

Coalition for Health (2023) Blueprint for Trustworthy AI Implementation Guidance and Assurance for Healthcare https://www.coalitionforhealthai.org/papers/blueprint-for-trustworthy-ai_V1.0.pdf

Davenport, Thomas, et Ravi Kalakota. The potential for artificial intelligence in healthcare Future Healthcare Journal 6, no 2 (juin 2019): 94-98. https://doi.org/10.7861/futurehosp.6-2-94.

Jones, E (2023) Explainer: What is a foundation model? Ada Lovelalce Institute 17 July 2023 https://www.adalovelaceinstitute.org/resource/foundation-models-explainer/

Harrison, Conrad J., et Chris J. Sidey-Gibbons. Machine learning in medicine: a practical introduction to natural language processing. BMC Medical Research Methodology 21, no 1 (31 juillet 2021): 158. https://doi.org/10.1186/s12874-021-01347-1.

Shaikhina, Torgyn, et Natalia A. Khovanova. Handling limited datasets with neural networks in medical applications: A small-data approach. Artificial Intelligence in Medicine 75 (1 janvier 2017): 51-63.

Sharon, T. (2018). When digital health meets digital capitalism, how many common goods are at stake? Big Data & Society, 5(2). https://doi.org/10.1177/2053951718819032

Sordo, Margarita. Introduction to neural networks in healthcare Consulté le 15 août 2023. https://www.academia.edu/20719514/Introduction_to_neural_networks_in_healthcare.

Sutton, Richard S., et Andrew G. Barto. Reinforcement learning: an introduction. Second edition. Adaptive computation and machine learning series. Cambridge, Massachusetts: The MIT Press, 2018.