Published on 16 October 2023

Fragmented Knowledge, Fragmented Body

Professor Joseph Sung
Dean, Lee Kong Chian School of Medicine

This year, I taught two sessions of Team-Based Learning (TBL): one on gastric diseases and one on liver conditions. TBL is an innovative way of teaching at the Lee Kong Chian School of Medicine. Instead of spoon-feeding medical students with lectures and hand-outs, we teach them how to learn by finding critical information for themselves, how to differentiate what is known and what is unknown, and how to critique medical literature on what is right and what is wrong. In our TBL classes, students are put into groups of six so that they can discuss the subject and work as a team; they can learn from each other and learn to help each other; they can debate and concur on what is the best option in patient management. Experts reason that this is a better way of teaching as it promotes active learning where one acquires and retains knowledge better. It also gives students a head start in learning to work in a team, with colleagues of different backgrounds and aptitude, and compensate for each other’s weakness. And they will view knowledge as an evolving body, to analyse information and challenge dogma. We at LKCMedicine know that it is an effective way of learning as students come out out this programme well informed and well equipped. 

During one of this year’s TBL sessions, however, I noticed some learning behaviour that worries me. It happened during a discussion on a case of chronic liver disease with portal hypertension. We put up a case in which a patient developed fever with leucocytosis, who then had a subsequent run into renal failure. Students were asked, what would be the cause of his febrile illness and kidney failure? Immediately, I saw students turning to their iPads and laptops to “consult” Google and ChatGPT. The answers they got obviously depended on how the question was phrased; the chatbot responded according to the information that was keyed in. Some said chest infection, some suggested peritonitis while others pointed to urinary tract infection. Most students, in fact a vast majority of them, turned to their search engine or generative AI to seek an answer for “liver disease with fever”, and without pondering carefully on the patient’s history, physical signs and blood chemistry, jumped straight into the diagnosis offered by machine. Of course, there could be more than one correct answer or diagnosis on why a patient with liver cirrhosis develops fever and renal failure. The answers given by the class shown on our statistics panel were “all-over-the-place”. I was actually not bothered by the diversity of opinions and answers. I was bothered by the way that they came to such conclusions and diagnoses. If that is the way how diagnosis will be made in the future, with no regard for clinical judgement, our profession is doomed. Imagine: a future when patients come to the clinic to present their illness and all their doctors do is ask ChatGPT on the diagnosis and next course of action.

Fever in liver cirrhosis can be due to infection in several sites: infected ascites in portal hypertension is a distinct possibility but aspiration pneumonia during hematemesis; septicemia as a result of a contaminated drip and thrombophlebitis; as well as urinary infection from a blocked Foley catheter should be considered. Spontaneous bacterial peritonitis is just one of many possibilities. It may even be a result of combined infection and inflammation at multiple sites in the body. On the other hand, renal failure can be a result of even more possibilities: nephrotoxic medications, dehydration, sepsis of any kind, hypovolemia, or a terminal complication of the liver disease-hepato-renal syndrome. However, when our students knock on the door of Dr Google, fever in cirrhosis is SBP; renal failure in cirrhosis is hepato-renal syndrome. If we don’t think about the other possibilities of infection, the diagnosis of SBP will mean sticking a needle into the belly of a patient with ascites. If we don’t think about other possibilities of renal failure, or checking the patient’s medicine, and monitoring blood pressure and urinary output carefully, by giving patients a challenge of intravascular fluid replacement, we are sentencing the patient to a terminal complication of hepatorenal syndrome.

There are several pitfalls in this way of “learning”. In fact, I should not even call it “learning”. When we consider the management of a patient’s condition by focusing on a single symptom or complaint, a fragment of the clinical presentation, and ignoring the other features and laboratory findings, it’s akin to playing the game of Blind Men and An Elephant. We are missing the forest by focusing too much on a single tree. The human body is very much compartmentalised and by taking this approach, diseases that affect multiple systems and variety of functions are overlooked. Furthermore, when we ask search engines and generative AI on a specific question and a specific answer is given, we forget that our body is complex and its organs are inter-connected. Our fragmented knowledge will not give us the full picture of what is going on and what needs to be fixed first. Think about complicated conditions such as diabetes mellitus, Crohn’s disease, lupus and immunodeficiency syndrome. We need to see the whole picture before coming to the major problem/s faced by our patients.

Searching for a simple answer to a direct question – that is not how medicine works. Importantly, I believe this is not the way we should be using technology and AI in Medicine. We should use technology wisely: to cut down mundane work and undertake detailed analysis of images, while not relying on them entirely. We should exercise discretion and make judgements, consider the physical and psychological aspects, to arrive at what’s best for our patients. In order to do so, we need the whole picture of normal anatomy, biochemistry and physiology, from vascular structure to neuronal network, from sugar metabolism to atherosclerosis, from microbiome ecosystem to specific infection. Because our body functions are connected, our gut-brain has an axis joining them, our organs work together as a system.

Remember, when laparoscopic surgery was first introduced, we were all very excited that keyhole surgery could achieve equally well what big wound laparotomy could do. But very soon, we realise that we cannot train surgeons just for laparoscopic surgery. They need to also know how to do a laparotomy in case laparoscopic operations fail or complications arise.

There is no short cut to success. We need to understand what is normal before we can discern what is abnormal. Our body is not fragmented – therefore our knowledge should not be fragmented.