Welcome to AI for Healthcare Professionals!
When you hear the term “AI,” you might think, “That has nothing to do with me,” or “It sounds too complicated.” However, in reality, AI has been utilized in medical practice and research for decades. Today, AI is closer to us than ever, supporting healthcare professionals in areas such as diagnostic assistance, image analysis, and improving the quality of treatment.
This blog aims to answer fundamental questions such as “What exactly is AI?” and “How is AI used in healthcare?” while also covering the latest trends and practical applications for medical professionals in an accessible manner.
A Brief Introduction
I am an internist involved in medical AI research at universities and engaged in the development of AI-powered digital medical devices at startups.
Yohsuke Takasaki, MD, PhD, ScM, MPA, MBA (Cantab)
After graduating from Okayama University Medical School, I pursued internal medicine and obtained a PhD in Medicine from the same university. I then earned a Master of Science from Harvard University, a Master of Public Administration from Columbia University, and an MBA from the University of Cambridge.
I have held various positions at the Ministry of Health, Labour and Welfare, overseeing medical DX, AI, medical device R&D, and healthcare policy. Additionally, I have experience in Japan’s IT sector and venture capital in the UK, focusing on business development and investment in healthcare. Currently, I serve as a professor at Okayama University, an associate at the University of Cambridge, and a visiting associate professor at Hiroshima University, engaging in academic research and education.
(For more details about my background and AI involvement, please refer to the end of this page.)
Why AI in Healthcare Matters
One of the things I constantly realize while working with AI is its rapid advancement. Groundbreaking research is published almost weekly, and new AI products are launched continuously. Keeping up with this information requires substantial effort, and without a foundational understanding of AI, it is easy to feel overwhelmed.
In fact, failing to grasp AI fundamentals now may result in an inability to comprehend its ever-expanding landscape in the future. In other words, now might be the last opportunity to learn AI from the basics.
Limited Resources for Healthcare AI Education
Despite the growing availability of AI-related books and articles, resources tailored for healthcare professionals remain scarce. Additionally, many of these resources are highly specialized, making it difficult for beginners to know where to start.
Even researchers eager to learn Python (a popular programming language for AI implementation) often find themselves struggling with the initial setup. Many feel discouraged by the technical barriers.
Purpose of This Blog
This “AI for Healthcare” blog serves as:
- An orientation for those new to AI.
- A platform for information exchange and discussion among AI users.
(A separate blog will focus on topics like AI utilization in medical settings, generative AI in healthcare, AI-driven innovation, and future AI predictions in medicine.)
We will start with “AI Basics You Can No Longer Ignore” from a medical professional’s perspective and gradually expand into the latest AI trends and specialized topics. My goal is to organize and share information while learning alongside you.
(I appreciate any corrections or suggestions to improve the accuracy of the content.)
Who Should Read This Blog?
- Clinicians: Looking for ways to integrate AI into their practice.
- Researchers: Exploring AI applications in their studies.
- Policy makers & public health professionals: Considering AI in policy-making and system development.
- Healthcare workers (doctors, nurses, pharmacists, lab technicians, administrators, etc.): Interested in leveraging AI to improve patient care.
Conclusion
AI is already making its way into medical practice, from diagnostics and treatment to research and administrative efficiency. Its potential continues to expand across all aspects of healthcare.
Through this blog, I will share essential AI knowledge, real-world applications, and the latest trends. Let’s explore and develop the world of healthcare AI together.
About My Journey with AI
For a detailed resume and professional background, please refer to my LinkedIn profile: https://www.linkedin.com/in/yohsuketakasaki
Here, I will focus on my experiences with AI.
Graduate School Days: Encounter with Social Medicine and Statistics
My journey into AI dates back to my graduate school days. I earned my PhD in medicine from the Department of Hygiene and Preventive Medicine (commonly classified as a social medicine), where the foundation of my studies was epidemiology and statistics. However, at the time, my understanding of statistics was limited to using it as a “tool,” and I did not fully grasp the theories and essence behind it.
Nevertheless, the knowledge of epidemiology and statistics that I acquired in graduate school laid the groundwork for my later entry into the world of AI. I believe that to deeply understand and master AI, three elements are essential:
① Statistics
② Mathematics
③ Computer Science (CS)
However, during my graduate school years, I could not have predicted the current advancements in AI, nor did I imagine that the statistics I was studying at the time would later be useful for learning AI.
By working with vast amounts of data and cultivating an approach to exploring causal relationships and risk assessments, I was able to prepare myself for expanding my studies into mathematics and computer science later on. My experience in graduate school was a valuable step in bridging medicine and AI.
Experience at the Ministry of Health, Labour and Welfare Japan: Between Research and Policy
After joining the Ministry of Health, Labour and Welfare (MHLW) Japan, I was involved in applying public health knowledge to policy development. Government officials have the opportunity to work in diverse fields, including other ministries, local governments, and overseas assignments, typically on a two-year rotation. I also took advantage of the ministry’s study abroad program and completed a master’s degree in social epidemiology at Harvard University and in economics at Columbia University.
At Columbia, I was struck by the extensive use of advanced mathematics and mathematical models in economics, far beyond what I had anticipated. In medical research, we typically conduct animal experiments or randomized controlled trials (RCTs) to eliminate biases and identify causal relationships. However, in economics, interventions such as “deliberately placing individuals in poverty” for experimental purposes are both ethically and practically impossible. Instead, economists construct mathematical models based on real-world data to explain complex social phenomena. This field relies far more on mathematics than medicine does, and I initially struggled with it.
By the end of my master’s program, however, I began to see similarities between medicine, which deals with the human body—a complex system of tens of trillions of interacting cells—and economics, which studies the interactions of billions of people. I was so captivated by this complexity that I even considered pursuing a PhD in economics. (I once took a PhD-level economics course at Columbia and was terrified of failing the course. To this day, I still have recurring nightmares about not earning enough credits and having to repeat a year…)
Risk Assessment at the Cabinet Office and a Reintroduction to Statistics
During my time as a government official, I worked at the Cabinet Office’s Food Safety Commission Secretariat, which scientifically evaluates food safety risks. Our main role was to assess how consuming hazardous substances in food affects human health and the probability of such effects occurring. This work involved gathering evidence from global animal studies and clinical research and quantitatively assessing risks to humans. The position felt like a hybrid between a scientist and a policymaker.
Through this experience, I was once again reminded of the importance of epidemiology and statistics, which led me to systematically relearn them. Around that time, I came across a statistics textbook (which I will introduce later) that was extremely easy to understand and deepened my appreciation for the subject.
Additionally, at the Food Safety Commission, I launched a new division that incorporated ICT, big data, and advanced statistical modeling into food safety risk assessment. As the head of this division, I gained practical experience in applying mathematical models and statistics—an invaluable experience that later facilitated my full-fledged study of AI.
Promoting Medical DX and AI at the Ministry of Health, Labour and Welfare and My Entrepreneurial Aspirations
At the MHLW, I served as the Director of the Emergency and Perinatal Medical Care Office, where I was involved in AI-driven policy planning for emergency medical services. Later, as the Director of the Medical IT Promotion Office, I worked on medical digital transformation (DX), specifically establishing data infrastructure to enable AI applications.
Through my work in advancing healthcare IT, DX, and AI from within the government, I gradually strengthened my resolve:
“In the second half of my career, I want to drive innovation in healthcare IT, DX, and AI as an entrepreneur and dedicate myself to making social security systems sustainable.”
After resigning from government service, I joined a major IT company, where I led new business development and investment strategies in the healthcare sector. I also had the opportunity to be seconded to a UK-based venture capital firm to study startup trends. Concurrently, with my company’s permission, I founded two startups focused on healthcare IT, DX, and AI, and I began seriously diving into AI studies.
Three Essential Elements for Learning AI
As mentioned earlier, I believe mastering AI requires three key elements:
① Statistics
② Mathematics
③ Computer Science (CS)
I had already acquired doctoral-level knowledge of statistics through my PhD and work experience, and I had a graduate-level understanding of mathematics from my economics studies. Therefore, I focused on learning computer science from the ground up.
For AI, the most challenging and time-consuming subject to master is statistics. The mathematics required to operate basic AI systems is at the high school level. However, to read research papers and develop new models, at least a university-level foundation in linear algebra, calculus, statistics, and optimization theory is necessary. To pursue cutting-edge research, advanced university or graduate-level knowledge—such as multivariate analysis, numerical optimization, stochastic processes, information theory, mathematical statistics, graph theory, and numerical analysis—is required.
For computer science, I started learning Python from scratch. Fortunately, I found an excellent learning resource (which I will introduce later) that allowed me to acquire skills efficiently. I realized that programming is best learned by building small, functional projects rather than trying to memorize syntax from the beginning. This approach helped sustain my motivation.
Challenges in Startups: Developing Innovative Medical AI Models
Since 2019, I have continued my studies and have now reached a level where I can create what I envision. While giving lectures and conducting research at universities, I am also working with our startup to develop digital medical devices aimed at saving the lives of mothers and babies. Additionally, we are tackling the creation of AI models so groundbreaking that I boldly describe them as “Nobel Prize-worthy.” While I must be cautious about publicly sharing details due to the nature of startups, we are actively publishing research papers and presenting at academic conferences.
Looking ahead, the era of explosive growth in medical and health data utilization is approaching, facilitated by innovations such as electronic health record-sharing services, nationwide medical information platforms, personal health record (PHR) systems, and regulatory frameworks for next-generation medical data usage. My nonprofit organization (a general incorporated association) is currently patenting a medical AI modeling method called The Formula of Life, based on time-series data analysis. By leveraging this, I aim to create concrete digital solutions from the ever-increasing wealth of medical data.
Although I have a long history in government service, I am committed to continuing my journey as a social entrepreneur, addressing societal and global challenges. I strongly encourage open innovation in The Formula of Life so that many others can benefit from it.
Future Prospects and a Message to Readers
Catching up with AI technology is challenging for any individual alone. I, too, have undergone countless trials and errors. However, there is great potential for healthcare professionals, passionate individuals tackling social security challenges with technology, and AI-driven problem solvers in government, startups, and large enterprises. I hope to support and collaborate with such individuals to drive transformation through AI.
Thank you for reading. I look forward to pioneering new possibilities in healthcare and AI together with you. Please feel free to reach out to me on LinkedIn or other social media if you have any thoughts or interests.
Let’s take this journey forward together!
コメント