This piece originally ran on Daily Nurse.
By Renee Hewitt, Editor & Content Strategist at Springer Publishing
Big data and machine learning already impact most aspects of modern life, so there is growing optimism about using artificial intelligence (AI) to transform health provider education.
Cristina Lussiana is a senior program manager of digital health and monitoring at Population Services International (PSI) and an expert in the health applications of AI.
Daily Nurse spoke with Lussiana about using AI to transform nursing education.
What follows is our interview, edited for length and clarity.
How has AI integration become more prevalent with technological advancements in the healthcare field?
Recently, progress in digital health has led to larger medical-related datasets, and technological advancements have made it possible for these larger datasets to be processed quickly. At the same time, there has been an increased awareness of AI’s potential to sustain and accelerate positive health outcomes. Think of electronic medical records (EMRs), medical imaging, and genomic data and how AI algorithms might identify patterns and trends faster and more accurately than human analysis. AI algorithms can quickly provide insights to help prevent, diagnose, and treat—leading to better health outcomes!
How can AI be used to improve health on a global scale?
In some countries, public health resources are scarce, and it’s mandatory to target these resources where they are more effective. This is where AI can play a role: helping to identify areas that we can invest in to build stronger and more resilient health systems. For example:
- Through personalized medicine based on an individual’s medical history and lifestyle, AI can support health practitioners in developing tailored screening and treatment plans. For example, in Kenya, we partnered with Audere on a research project to assess HealthPulse AI, a suite of AI-powered tools for clinicians, Community Health Workers (CHWs), and health consumers. We studied whether the tools could improve the accuracy of administering and interpreting malaria Rapid Diagnostic Test kits (RDTs) by CHWs and health workers in private clinics. HealthPulse AI uses machine learning and computer vision to improve the accuracy of rapid diagnostic test kit results. It requires only an image of the RDT captured by the user’s smartphone to interpret the results of the RDT and can read even the faint test result lines that expert test readers may miss. This pilot project demonstrated that AI-powered tools in health workers’ hands and CHWs’ hands could improve the accuracy and interpretation of rapid diagnostic tests and positively impact the quality of care consumers receive. Additionally, it holds potential as a mobile tool that can be scaled up for use in low-resource settings with potential benefits as a supportive supervision, diagnostic, and surveillance tool. The project also confirmed that health facilities and CHWs would accept and welcome such a tool.
- When accessing health-related information, AI-powered chatbots can provide accurate information and direct consumers to relevant health facilities, even in areas with limited resources. For example, in late 2022, PSI and Babylon launched AI OI, a new digital health service focused on supporting people in low-income communities to make informed decisions about their health and efficiently navigate the healthcare system. Piloted in Vietnam, the new, free 24/7 service gives users more control over their access to healthcare, triages users to the appropriate level of care, and signposts to high-quality providers in their local area. And it saves people time and subsequent loss of income from taking time off work and paying unnecessary out-of-pocket expenses.
- By analyzing large datasets and identifying patterns, AI algorithms can support governments and public health implementers in predicting outbreaks, playing an important function in health security.
- For research and development, AI has the potential to advance progress in medicine by, for example, identifying potential drug targets and improving the sensitivity and specificity of diagnosis.
Yes, AI can potentially improve health globally – but only once we have determined who leads and how we implement it.
Does ChatGPT provide nursing students with a convenient and accessible way to access information and practice clinical skills? Talk about the potential for ChatGPT-assisted learning in nursing education.
ChatGPT provides students with a wealth of information that can fit their educational needs.
- ChatGPT offers students a variety of case scenarios and medical questions for research and practice.
- ChatGPT can also facilitate group work – students can work on similar case scenarios and exchange ideas.
- Students can stay current on the latest guidelines and best practices in medicine and have access to nursing education materials for free, even from low-resource areas.
But it does not, and should not, replace in-person training. With ChatGPT, students won’t have the in-person experience to treat patients, including emotional intelligence development and hands-on clinical care.
What are the potential misuses of ChatGPT in nursing education?
ChatGPT can be a valuable tool in nursing education; however, there are some misuses that students need to be aware of, namely:
- AI-related bias: AI algorithms are based on the type of data they are fed. Despite technology advancements to ensure a great deal of variety in datasets, there are limitations to the accuracy and usefulness of AI technology in contexts and scenarios that represent a case for a bias, resulting in inaccurate or unfair recommendations.
- Lack of domain-specific knowledge: In some cases, ChatGPT might not be up to date on emerging diseases and changing patterns in specific health-related topics or specialized medical fields.
- Lack of or limited understanding of context: This might quickly change even when trained in a specific context. ChatGPT might not always correctly understand the context where its answers should operate, resulting in inappropriate responses.
- Overreliance on technology: Similar to how calculators can limit mental math abilities, there is the risk that ChatGPT users over-rely on the technology and don’t train their clinical memory and critical thinking muscles. These are skills for nurses to make decisions adequate to their patient’s unique needs and contexts.
As AI continues to develop, will it replace educators in the future?
According to Oxford Languages, education can be defined as an enlightening experience. This goes beyond transferring knowledge and skills from one human being (or an AI) to another. AI technology cannot offer the level of human interaction, debate, discussion, and involvement needed to spark a rich learning experience. Also, because of the limitations of AI technology listed above, it is unlikely that the role of educators will be replaced by AI technology. AI has the potential to complement and enhance education. Still, it cannot replace what educators offer regarding human interaction and soft skills like critical thinking, understanding of the context, empathy, communication skills, and creativity.
If acquiring knowledge becomes increasingly accessible, what will be the role of higher education in an AI world?
Assuming that in the next ten years, knowledge will be increasingly accessible through AI technology for free, the role of higher education might evolve to concentrate on skills and competencies that cannot be learned via AI, like emotional intelligence, critical thinking, interdisciplinary learning, contextualization, and prioritization. Furthermore, highly specialized education in medical fields requires education that is not AI-based.
What kind of education and training do we need to provide our nursing students so they become highly qualified geriatric nursing professionals?
The world population is getting older, and we know this trend won’t reverse. Hence, nursing needs highly qualified professionals who confidently and competently manage geriatric patients. This entails offering students a comprehensive package of educational knowledge and skills that prepare them to care for an aging population in different aspects, from the medical to the mental one. Specific education topics include geriatric diseases, management of chronic diseases, mental health, palliative care, emotional intelligence, and ethics. This ensures geriatric nursing professionals have diverse skills that equip them to deal with older adults who might experience chronic or aging diseases, mental health issues, and emotional burdens related to loneliness, isolation, and anxiety due to a fast-changing world where it’s difficult to catch up. Furthermore, we want to ensure we equip geriatric nursing professionals with the tools they need to support older adults in making life/death choices in an ethical way.
Does this mean that training critical thinking, rather than delivering the content of the class materials, becomes more important than ever?
Absolutely, and this is not just about AI! Knowledge and information have been broadly available and accessible since the world wide web. Now, AI technology presents this information in a tailored and targeted way. However, the need to develop critical thinking and emotional intelligence is still there. Problem-solving, debating, collaboration, coordination, prioritization, evidence-based decision-making, making assumptions, and contextualization are soft skills that allow students to apply knowledge and information, regardless of where they have acquired these from (classroom training, AI chats, etc.).
How should we evaluate students’ competency in acquiring knowledge in the classroom? Should we start implementing traditional paper and pencil formats for exams?
Traditional paper and pencil formats are still very effective for evaluating students’ competency. These can be complemented by other assessment methods like practical skills assessments, where students are required to prove their competency on a case scenario (real or simulated), or project-based assessments, where students are required to develop a specific project to demonstrate the knowledge and skills acquired (i.e., a literature review, a group project, etc.).
While we are making concerted efforts to address health disparities and promote equity globally, do you see the advancement of technology further widening the digital divide across populations with various socio-demographic characteristics (e.g., age, socioeconomic status, and geographic locations)?
Digital health and technology, in general, have the potential to improve healthcare even in low-resource settings. However, it is widely accepted and recognized that technological advancements can widen the digital divide across socio-demographic groups because not all groups have equal access to technology. There is the risk that technological advancements benefit only some of us, leaving behind people with specific socio-demographic characteristics.
There are disparities in the data used to train AI algorithms because underserved populations often lack access to digital health, and their data is not represented in these datasets. This can result in health disparities because AI algorithms do not represent diverse populations, leading to inaccurate or discriminatory results, particularly for those living in low-and-middle-income countries.
As we work toward Universal Health Coverage, it is important to design technology in a way that takes equity and inclusion into consideration, and that is freely available and accessible to all population segments, like zero-rate or low-cost internet access, digital training, language-inclusive content, inclusion of technology elements for people with disabilities, and so on.
Anything else to add?
We now collect data while we offer services like diagnostic, triaging, and signposting. That is because of Fast Healthcare Interoperability Resources (FHIR). This standard describes how to collect, store, use, exchange, and distribute workforce-related information and patient and health data, regardless of the application used. By implementing FHIR and consumer-facing data, we can identify patterns in digital health that point to causalities the human brain cannot quickly identify.
