Learning to code a decade ago revolutionized my approach to medicine. By developing healthcare apps, I have managed to help many healthcare departments or units solve work-related issues. The best part? I achieved these while still practising full-time as a clinician. Coding empowers doctors to tackle real-world challenges in healthcare. Whether it is inefficient software, gaps in patient care coordination, or the ever-growing volume of medical data, coding equips us with the tools to make a difference. Do not be discouraged if you do not have a computer science degree – there are fantastic resources designed for busy professionals to acquire coding skills one step at a time. Why Should Doctors Learn to Code? The short answer is: being able to directly apply your clinical expertise in technology. Such flexibility can be very powerful. As a doctor-coder, some tech products that you can work on include: Skills Transfer Seamlessly: Why You’re Already Well-Equipped My experience practicing medicine alongside collaborating with programmers has highlighted the remarkable overlap in the way we (i.e. doctors and programmers) think. Here are a few core skills doctors possess that give them a natural edge in becoming skilled programmers: Getting Started with Your Coding Journey Explore the numerous online resources and books available – find one that matches your learning style. I favour project-based learning over traditional chapter-by-chapter methods for a more engaging experience. The rewarding experience that you feel when you are making progress on a project, can help you to maintain your motivation to keep going. How should you get started? Let me share some tips that I hope can help you in your coding journey: 1. Choosing Your First Programming Language There are just too many programming languages out there and choosing one can be overwhelming for a beginner. I think there are a few excellent starting points: Once you have mastered one programming language, learning additional ones becomes significantly easier. This is because most languages share common patterns, concepts, and keywords. As you gain fluency in multiple languages, you will start to appreciate the design decisions behind them and understand why specific languages are better suited for particular tasks. 2. Online Learning Platforms Online learning platforms are fantastic resources for those want to learn programming because of their interactive nature and the contents are often spread into small chunks, allowing you to follow along at your own pace, one small topic at a time. If you prefer online learning, try to browse these: 3. Use AI to Enhance Your Learning Journey The emergence of powerful language models like ChatGPT and Gemini has transformed the way we learn. I agree that, used responsibly, AI can become a valuable learning companion. Its ability to synthesize massive datasets makes finding resources and solving coding challenges quicker and more efficient. Personally, I have saved countless hours using AI to clarify concepts and troubleshoot code, rather than spending endless time on search engines looking for answers. 4. Practice consistently As a fellow clinician, I understand how busy our schedules are. However, consistent practice is key to making progress with coding. Instead of aimlessly scrolling social media during quick breaks, turn those into mini-coding sessions. Whether it is 10 minutes of refining your project or watching a coding tutorial, consistency is essential. The more time you dedicate, the faster you will see results! Conclusion Coding is not just about technology; it is about unlocking a physician’s potential to create solutions that directly impact patient lives. Consider the challenges you see in your practice – chances are, code could help solve them. Start small, explore the vast array of learning resources, and join a community of like-minded doctors-turned-coders. The future of healthcare innovation might just be written by you.

In this final part of the AI in Medicine article series, we will explore the transformative potential of AI and its role in shaping the future of healthcare. The integration of AI into medicine is no longer science fiction. It is happening right now, reshaping how doctors diagnose, treat, and interact with patients. While incredibly exciting, this transformation also raises important questions and responsibilities. Let’s explore some key themes at the forefront of the AI revolution in healthcare. The AI-Doctor Partnership: A New Era AI is not going to replace doctors. Instead, it is ushering in a new era defined by intelligent collaboration. Consider a scenario where you are examining X-ray images for signs of a subtle fracture. A dedicated radiology AI can simultaneously analyze the images and highlight areas of concern that might have been missed by the human eye. This is the doctor-AI partnership at its best – complementary skills, working in tandem to provide better patient care. Just as we use electronic BP and pulse monitoring for accuracy, we now use AI medical assistants to augment our senses, which—let’s admit it—are error-prone, especially when we are tired from working long hours. With routine tasks streamlined and complex information readily summarized by AI, doctors can finally reclaim something precious: time. It is time to focus on the irreplaceable aspects of medicine – to build rapport with patients, provide empathy, and tailor treatment plans to the unique person beyond their chart. The Exciting Possibilities of AI AI in medicine is still in its early stages and has a long way to go, but the potential is transformative. For example: Addressing Ethical Concerns: Building Trust The benefits of AI in medicine are clear, but to gain widespread acceptance and ensure proper use, we need to address the ethical considerations.  The ethical issues surrounding AI have been predicted for many years, even before the current AI revolution. However, with AI use becoming widespread, we need to start seriously considering these implications. Here are some questions we should ask when discussing AI ethics in the context of healthcare: The Human Touch Remains Essential Amidst the exhilarating array of medical possibilities, it is crucial to acknowledge that the human element should be maintained as the core foundation of medicine. AI can crunch data and offer insights, but it cannot replace the doctor-patient connection. Empathy, compassion, and the ability to understand the nuances of a complex health situation are uniquely human skills, and I doubt any machine can ever reach this state. What sets humans apart is a lifetime of experiences, which a machine can never organically acquire. A machine, after all, is merely a machine, processing binary digits. It cannot experience the genuine emotions and connections that humans do. The greatest successes in AI medicine will come from finding the perfect balance between technological innovation and human intuition. The Evolving Doctor-AI Relationship and Continuous Learning The rapid advancement of AI requires doctors to evolve their interactions with these tools as well.  Professional and ethical AI use can be life-saving. Therefore, ongoing education about its use in medicine will be vital for doctors to use it effectively. As the integration of digital technology into clinical medicine may soon become increasingly evident and beneficial for improving healthcare, we may start seeing doctors attending not only continuous medical education (CMEs) but also continuous technology education (CTEs) to keep up with these advancements. By embracing AI responsibly, ethically, and always with patient well-being as the guiding principle, we usher in a future where the possibilities for better, smarter, and more compassionate healthcare are truly limitless.

In this third part of our AI in healthcare series, we will examine prompting techniques—a skill essential to utilizing Large Language Models effectively. You have seen the amazing capabilities of your new AI assistant, but how do you unlock its full potential? Imagine it like mastering any powerful new clinical tool or technique – a little practice goes a long way. Here is a deep dive into the art of interacting with AI to maximize its effectiveness in your daily practice. Exploring Your AI Toolkit: A Look at Available Options The world of AI is still relatively new and constantly evolving, but there is already a wealth of tools available right now. There are many tools designed specifically for healthcare professionals. Here is a breakdown of its levels (do note that the word ‘AI’ in this article refers to LLMs until stated otherwise): Free or Paid Options? Many AI tools offer basic functionality for free, allowing you to experiment and assess their capabilities before committing. These “freemium” options are a great way to get started and familiarize yourself with AI’s potential in your workflow. As your comfort level increases, you might consider paid tiers that offer advanced features and potentially higher levels of accuracy and reliability. Remember, paid-for AI products do not necessarily mean better – thoroughly evaluate any tool before integrating it into your practice. In most cases, even the free versions can be more than adequate for your daily use cases. The Art of the Prompt The way you ask questions has a dramatic impact on the information your AI assistant retrieves. Forget long, convoluted sentences. Think of your AI as a highly intelligent but slightly inexperienced intern. Clear, direct questions will get you the best results. Here is an example: Think of typing prompts like using a search engine more interactively. When we use search engines, vague queries lead to a tangled mess of results, while focused keywords deliver exactly what you need. The same principles apply when typing prompts. Here are some additional tips for crafting effective prompts about clinical issues and topics: Beyond Simple Questions: Exploring the Power of Advanced Prompts While basic questions are valuable, AI can handle more sophisticated requests. Imagine you have a complex patient case with a mix of symptoms that do not seem to fit a clear diagnosis. You can use your AI assistant to explore possibilities: These advanced strategies can unlock the true potential of AI as a clinical and research assistant, helping you stay current on the latest medical advancements and offer personalized treatment options for your patients. A Critical Eye: Building AI Literacy for Doctors AI is a powerful tool, but it is not infallible. Responsible use hinges on doctors having a basic understanding of these tools and their limitations, which include: In our next and final part of this AI in medicine series, we will look ahead to the future of AI in medicine. We will discuss exciting possibilities, ethical considerations, and the evolving doctor-AI partnership in a world of ever-smarter machines.

In the previous instalment of this series, we delved into the origins of AI, exploring its development and the unparalleled impact of recent advancements like GPT-based LLMs on the AI landscape. Now, let us shift our focus to healthcare, the industry we, as doctors, are deeply invested in. We shall examine how AI is revolutionising this critical sector. Remember those bulky medical textbooks and endless stacks of patient files? Imagine an AI that can find the exact information you need within seconds. Add in the ability to draft routine paperwork and pinpoint subtle test results patterns. As doctors, wouldn’t this speed up our workflow significantly? This is just one of the many examples to appreciate the potential of AI in medicine. Before we go into panic mode by thinking AI will take over our jobs, I invite all of us to reframe our understanding of what AI does. I am convinced AI is not here to take your job. Instead, they are powerful tools to help you work better and more efficiently. The combination of AI and skilled professionals is often better than working without AI. Let me give some specific examples. AI as Your Clinical Research Assistant The ever-expanding ocean of medical knowledge can be daunting. New studies are published daily, drug guidelines evolve, and keeping pace can feel like a never-ending race. Here is where your AI research assistant comes to the rescue.  Imagine a scenario where you need a quick update on the latest clinical research for a specific treatment option for lung cancer. You can ask the AI (e.g. ChatGPT) by typing in this instruction: “Summarize recent studies on the use of immunotherapy for advanced lung cancer”. You will get a concise report with key findings in an easy-to-digest format. Struggling to decipher a complex medical paper on a rare condition? AI can translate the jargon into a clear and accessible summary, allowing you to focus on the practical implications for your patients.  This information overload is not just about research papers. Keeping track of a patient’s specific medical history, allergies, and medication interactions can be challenging, especially with the often overly-used shorthand notations and jargon that even fellow doctors struggle to understand. AI can help you analyze a patient’s electronic health record (EHR) and highlight potentially relevant details, saving you valuable time and ensuring you have a comprehensive picture before making treatment decisions. To achieve this, you can try this prompt: “Summarize the patient’s illness, including a detailed description of symptoms, diagnostic procedures conducted, treatments administered and how the patient responded over time.” AI as Your Clinical Administrative Assistant Doctors spend a significant portion of their time on administrative tasks, time that could be better spent with patients. AI can be your secret weapon for reclaiming those lost hours. Imagine dictating your notes during a patient encounter and having AI automatically generate a clear and concise consultation report, ready for your review and edits. It takes the tedious typing out of the equation and allows you to capture your thoughts while still fresh. Need a basic discharge instruction sheet for a common condition? Delegate that to your AI assistant too. Simply provide the necessary details like diagnosis and medication information, and the AI can generate a draft that you can personalise. This frees you up to focus on more complex cases or spend extra time with patients who need additional explanation. If you work in patient education, you might need to an endless stream of patient emails with basic questions. Here too, AI can be a lifesaver. Train your AI assistant to handle routine inquiries about medication side effects, appointment scheduling, or post-operative care instructions. It can even answer frequently asked questions and triage more complex concerns directly to you, ensuring that urgent matters get your immediate attention. Just remember to audit the AI output frequently to ensure it sticks to the guidelines and protocols practised in your centre. AI as The Patient’s Trend Tracker AI excels at analysing vast amounts of data and identifying subtle connections that might escape even the most experienced doctor. Imagine a scenario where a patient’s blood test results appear relatively normal at first glance. Your AI assistant, however, highlights a potential concern: a slight elevation in a specific enzyme level, combined with a seemingly unrelated medication the patient is taking.  This could flag a rare drug interaction, allowing you to investigate further and potentially prevent a serious adverse reaction. The power of AI goes beyond single test results. Over time, your AI assistant can learn to analyse a patient’s entire medical history, identifying trends or patterns that might suggest an emerging health issue. For instance, it might detect a gradual decline in kidney function based on past lab reports, prompting you to order further tests and intervene early. This is no different than using AI to analyse data patterns in stock markets, business, road traffic and other similar use cases. We are incorporating the same mechanism in our workflow when managing patients. It is like having an extra set of eyes that never gets tired and can monitor trends over long periods. AI is a Tool to Empower, Not to Replace It is crucial to remember that AI, however powerful, should remain as a tool. It augments your expertise, not replaces it. Think of AI systems as just the latest medical instrument or clinical tool, much like a stethoscope or an advanced imaging system when they were first introduced. It provides valuable insights, but the diagnosis, treatment plan, and ultimately, the patient care decisions remain in your capable hands. The most successful doctor-AI partnerships will be built on a foundation of trust and understanding.  It is about recognizing the strengths and limitations of this technology and harnessing its potential to elevate the quality of care you provide. In the next part, we will delve into the practicalities of using AI. We will explore the art of crafting effective prompts (giving clear instructions to your AI) and introduce you

In this four-part series of articles, we will take a closer look at AI and its potential applications in healthcare. To begin, let us delve into a concise history of AI to provide context for its current and future uses in this field. Imagine a world where computers can think like we do, or at least seem to! That is the dream behind Artificial Intelligence (AI), which unlike many would think, is a journey that began long before our smartphones and self-driving cars. The Early Days of AI The idea of intelligent machines has been around for centuries, from myths of animated statues to science fiction stories about robot companions. But it was not until the rise of computers in the mid-20th century that these dreams started to feel possible. In 1950, a mathematician and computer scientist Alan Turing pondered a fundamental question: could a machine ever fool a human into thinking it was one of us? He devised the “Turing Test,” which is a benchmark for measuring machine intelligence that continues to spark debate today.  During the same era, as chips and computers became increasingly available and more widely used, researchers sought to develop computer programs capable of playing games like chess. I vividly recall reading about the legendary battle between chess grandmaster Garry Kasparov and the Deep Blue Machine when I was younger. Such games demanded strategic thinking and decision-making, abilities once considered exclusively human. As video games gained popularity, programmers endeavoured to create AI opponents that could challenge human players. Some AI opponents were programmed to be challenging, while others were intended to be easier to defeat. Often, these video games allow us to choose AI opponent difficulty settings ranging from easy to hard. Beyond games, machines began making programmatic suggestions to us when we use the internet, such as targeted advertisements and recommended articles and videos. Developers can code a simple recommendation algorithm which tracks the last video that we watched, fetches its title and tags, and uses these as keywords to display recommended videos that have the same tags or almost similar titles. These achievements hinted at a future where computers would follow instructions and make their own choices, representing a significant shift in our understanding of machine capabilities. Machine Learning: Setting The Scene for AI Revolution As mentioned earlier, early AI often involved explicitly programming rules and knowledge into a machine. Think of it like a complex flowchart: “IF a patient has symptom X, THEN consider disease Y.” But we know such a rigid algorithmic approach became unwieldy as medical knowledge grew. Enter machine learning, a fundamental shift in AI development. Instead of hard-coded rules, machine learning algorithms, or called models, are fed massive amounts of data. These models discover patterns and connections on their own, sometimes even surprising their creators. This is the magic behind your email’s spam filter, your streaming service knowing what show to suggest, and those uncannily targeted ads that pop up in your web browser. Before machine learning was born, machines could perform tasks and make suggestions based on rules programmers created. But that was a pain — it took forever and needed constant updates. Machine learning changed everything. Now, machines are not slaves to instructions. They can “think” on their own. These sophisticated algorithms let machines learn from data, see patterns, and decide what to do. Machines can now give answers that are just as good or even better than the ones if we used to code ourselves. The release of ChatGPT in early 2023 marked the beginning of an era that many experts have hailed as the next industrial revolution. ChatGPT is an AI product that utilizes vast amounts of text (language), enabling it to generate text based on prompts provided by the user. Consequently, this model is also known as a large language model (LLM). Developers employ various machine-learning algorithms to create LLMs. ChatGPT utilizes a form of model known as a generative pre-trained transformer, or GPT for short. Apart from ChatGPT, there are also other popular LLMs such as Gemini, Claude, Llama and many others. Prompting, defined as the act of communicating with these LLMs, is fast becoming an important skill. We will explore key prompting skills in subsequent parts of this four-part article. Talking Assistants, Self-Driving Cars, Robotic Surgeries and Beyond Over the last decade or so, AI has woven its way into the fabric of our daily lives. Even before LLMs came to become mainstream, “old-school” virtual assistants like Siri and Alexa understood our requests with surprising accuracy. From facial recognition in our phones to the algorithms that personalize our social media feeds, AI is subtly shaping our experiences. Its impact extends well beyond consumer gadgets. Self-driving cars, systems that detect financial fraud, and AI-powered robots assisting in complex surgeries are just a few examples of how AI is transforming industries and challenging our preconceptions about what machines can do. The Brains Behind AI Beyond LLMs, AI development is a melting pot of disciplines, attracting experts from diverse backgrounds: AI is still a young and rapidly evolving field. However, we cannot deny the scare that was caused since the release of the groundbreaking ChatGPT. This led to questions about bias and the potential impact on jobs, raising important challenges for society. Nevertheless, its potential to revolutionise how we live and work is undeniable. I believe it is best to position ourselves without overly relying on AI or completely denying its benefits. Taking the moderate approach is the way to go. In the next part of this series, we will dive into how the AI revolution is transforming the world of medicine.