AI helps predict diseases more accurately and personalize treatment, according to Professor Karin Verspoor, Head of Computer Technology at RMIT University Australia.
The application of AI in healthcare is gradually becoming a global trend. Discussing this issue, Professor Karin Verspoor had a conversation with VnExpress about the advancements in this field.
Professor, could you please analyze the global trends in AI applications in healthcare?
- In healthcare, one of the areas where AI is most fully utilized is image processing. For example, computer vision applications leverage machine learning to diagnose and detect diseases. This technology is also used to interpret chest X-ray results or classify skin lesions that may be signs of cancer. Many hospitals also deploy highly accurate and efficient robotic surgical assistants, supporting surgeons based on the characteristics of each patient.
We are witnessing increasing progress in using AI to guide clinical decision-making by utilizing a wider variety of clinical data. For example, data recorded in electronic health records – including both structured data (e.g., blood biomarkers, vital signs) and unstructured data (data from notes, reports, genetic information).
Another notable trend is the broader application of AI in clinical settings. You might see devices that can assist in building clinical documentation through automated note-taking, clinical narration during surgery, or recording patient history during doctor consultations.
Professor Karin Verspoor at the third annual Vietnam-Australia seminar on cooperation to promote Industry 4.0, held on October 20, 2022. Photo: Provided by the interviewee .
- What are the reasons why image processing technology has such a profound impact on the healthcare field?
This stems from the unique characteristics of the healthcare industry, which has more frequent and systematic data than many other types of clinical data. In the healthcare sector, there is a limited number of imaging devices and manufacturers, so the data is quite consistent.
Furthermore, images are well-suited to current AI algorithms. They can be viewed as a completely dense pixel matrix, meaning every cell in the matrix has a value. This type of data is well-suited to the mathematical presentation and analysis methods that AI approaches can perform.
Additionally, there is a large amount of labeled image data – for example, known diagnoses associated with each image. This means that supervised machine learning implementation is very simple. These systems have proven highly effective and perform as well as, or even better than, human experts in some cases.
- Within that overall picture, how is AI being applied in public healthcare in Vietnam?
- In developing countries, the deployment of software systems such as electronic health records may be less common. These countries also have less access to technology and healthcare resources, impacting the development of certain applications that rely on electronic data collection.
However, technology and AI can still bring significant benefits to users in these countries and also in Vietnam. AI provides a wealth of specialized knowledge even when it is not readily available locally. Instead of specialized equipment, you can use sensors on common products such as mobile phones and smartwatches to record health data. Some tools can analyze recorded cough sounds to diagnose Covid-19 or detect atrial fibrillation from heart rate using data from these devices.
Smart health assistants can be deployed through an app, thereby empowering patients with greater control over their health.
So what are the barriers to applying AI in the healthcare sector?
- The main barrier to AI in clinical decision-making relates to data collection on the Vietnamese population. Any AI tool needs to be tailored to the specific characteristics of the population. This means the input data must be consistent with the data the model was trained on.
AI tools are often not easily transferred from one context to another. This implies that, for AI to confidently function well in the Vietnamese context, the tools need to be adapted and evaluated appropriately within that context. This requires investment in digital infrastructure in Vietnam's healthcare facilities. This investment should be balanced across all aspects: healthcare facilities, electronic health record systems, and mechanisms for sharing and linking data between healthcare providers.
A greater challenge involves identifying the specific issues that need attention within Vietnam's unique environment to maximize AI's value. This requires collaboration among researchers, AI innovators, and healthcare leaders to identify opportunities and prioritize investments.
- Could you share some experiences from Australia in this field?
In Australia , Covid-19 spurred the adoption of digital health technologies, intensifying demand. Lockdowns and restrictions led people to seek online healthcare. This transformed the healthcare landscape, shaping a trend towards using technology to support healthcare and overall well-being.
These changes have garnered attention and support from the community, leading to national dialogues—within government and in the media—about regulation of software as a medical device, the ethics of using AI in a healthcare context, and the value of medical data as a public resource. Along with its value, entities need to respect the sensitivity and privacy of this data.
I think Vietnam could learn from this experience, which is to engage the public and help them understand the opportunities that AI offers in healthcare. Ultimately, it is patients and consumers who will benefit from the adoption of these technologies. But we will also rely on their data to build and evaluate them. Therefore, it is crucial to build trust in AI systems, both from patients and healthcare providers.
Professor Karin Verspoor (far left) during a discussion with experts on the potential of AI. Photo: Provided by the subject .
- What are your predictions regarding the future development of AI in healthcare?
Today, AI is more present in people's minds than ever before. The excitement surrounding ChatGPT and AI generation has led to increasing user interest in using artificial intelligence to solve countless problems in business and life.
The application of AI to health and well-being is no exception, and we will certainly see increasing innovation in this field. I believe there will be many opportunities to leverage AI to improve patient healthcare through the integration of multimodal data and sophisticated predictive models.
AI helps predict patient outcomes and disease progression more accurately, and provides highly personalized treatment plans. We will be able to leverage the technology to capture recorded medical activity, thereby providing knowledge and evidence about the impact of treatment. This leads to further improvements in practice – an ethical cycle known as a Learning Health System.
We can improve the patient experience by proactively suggesting steps in the treatment process, providing clinicians with relevant information to support their decisions. We can even enhance the patient experience by leveraging AI to make interactions with the healthcare system more "human." For example, assisting with preparation tasks and documentation to free up doctors and give them more time to talk to their patients. Some real-time translation tools allow for multilingual settings, helping to translate complex medical terminology into more understandable information, increasing the effectiveness of patient communication.
Patients will have more autonomy in their own healthcare. They will also use digital technology to collect, manage, analyze, and interpret their own health data, thereby gaining more information when interacting with the healthcare system.
Minh Tu
Source link
Comment (0)