Quick Summary

• Developments in Artificial Intelligence (AI) models have seen an exponential rise in recent times making them more accessible.

• AI technologies such as Machine Learning (ML), Deep Learning (DL), Computer Vision (CV) and Natural Language Processing (NLP) can be used to augment and assist clinicians in providing higher-quality care for patients.

• CV technologies allow for more accurate diagnostic radiology

• NLP and data analytic systems can increase efficiency in medical documentation workflows and medical research.

• Improved deep learning algorithms have the potential to revolutionise anaesthesia delivery and monitoring facilitating better decision making and surgical outcomes.

• AI can also improve medical education and curricula by creating more engaging content and personalised and targeted feedback for students.

• Despite the benefits, caution should be exercised in AI’s implementation into healthcare due to many ethical questions regarding patient privacy, data security, algorithmic bias, and the potential displacement of healthcare professionals that exist.

Preamble

In the late 1900s and early 2000s, we witnessed an explosion in knowledge and technological advancements brought about by the internet, most of which have become irreplaceable aspects of daily life we take for granted today. A technological revolution seems to be in motion once again: with the recent meteoric rise of Artificial Intelligence (AI) models, is our society at another turning point in our evolution? Should we adapt to and embrace the possibilities brought about by AI, or shoulqd we exercise greater caution in fear of experiencing the Singularity?

Many of today’s technological giants including Apple (Apple Intelligence), Google (Google Gemini), and Microsoft (Copilot+ PC) believe it is time we embrace these changes and explore the opportunities created by integrating AI into our daily lives. Perhaps it is time that we follow suit and consider how the field of medicine can benefit from these new technologies. What better way to begin, than to ask ChatGPT what AI has to offer medicine?

[ChatGPT: AI is like the future’s secret weapon in scrubs. It’s crunching numbers faster than a caffeine-fueled resident, spotting patterns, and predicting outcomes like a medical oracle. Plus, it’s the ultimate multitasker—handling everything from paperwork to breakthrough research faster than you can say “robotic surgeon.” Basically, AI is here to make medicine smarter, snappier, and surprisingly hilarious.]

AI in Medicine

Although we may not be aware, AI has been utilised in medicine since the 1960’s with the development of our familiar Medical Literature Analysis and Retrieval System (MEDLINE) and the web-based search engine PubMed by the National Library of Medicine. Its progress stagnated during the “AI Winter” from the 1970’s to the late 1990’s, largely attributed to the excessive cost of developing and maintaining digital information databases and the perceived limitations of AI. Fast forward to more present times, our increased understanding of the AI field has allowed for the exponential development of new technologies with untapped potential in modern clinical medicine.

To delve into these novel technologies, we must first understand some basic concepts of AI.

• Machine Learning (ML) enables machines to learn from data patterns, identify trends, and make predictions.

• Deep Learning (DL), a more advanced form of ML, aims to mimic the intricate and simultaneous processing ability of the human brain. It utilises artificial neural networks to analyse vast datasets to make multifactorial decisions. It also has the ability to adapt, as it does not use a set number of traits that require human input like in simple ML algorithms. This allows for customisation of patient care as it self-formulates more dynamic and personalised approaches to management.

• Computer Vision (CV) allows computers to “see” and interpret images and videos. This is revolutionising medical imaging, enabling rapid and systematic interpretations to assist in diagnosis.

• Natural Language Processing (NLP) allows computers to understand and process human language, extracting information from medical textbooks, patient records, and research papers.

Utilising CV technologies, medical imaging has been the poster child of the adoption of AI in medicine. With its application being shown to detect lesions often missed by the human eye and create differential diagnoses, it can improve accuracy, consistency, and efficiency in reporting. Arterys, the first U.S. Food and Drug Administration (FDA) approved clinical cloud-based DL application in health care, developed CardioAI. This program analysed cardiac MRIs in a matter of seconds, providing information such as cardiac ejection fraction, and the application has since expanded to include liver and lung imaging for oncologic investigations, chest and musculoskeletal x-rays, and non-contrast head CT images. CV has also been used outside the radiologic realm with applications in skin cancer diagnostics showcasing its ability to differentiate non-melanoma and melanoma skin cancers with accuracy comparable to that of experts. Further, some AI models have even been shown to improve accuracy in chronic disease screening with more accurate cardiovascular risk predictions when compared with the established algorithm and guidelines defined by the American College of Cardiology.

Outside of its diagnostic capabilities, AI may be able to improve efficiency and workflows, thereby reducing healthcare staff workload and positively impacting physician mental health in an ever-resource-limited healthcare environment. An “AI resident” created by Australia’s very own Heidi Health, aims to do this by using NLP techniques to produce structured clinic notes, discharge documentation and specialist summary letters from voice recordings of the patient interaction. AI can also be used to streamline patient intake by reducing wait times. MANDY, a mobile-based primary care chatbot application developed at the University of Auckland, aims to achieve this by allowing patients to input their medical history and describe their symptoms in their own words. This information is then analysed to create a report for doctors – very similar to how a medical student or junior doctor seeks input from seniors after clerking a patient.

AI in Anaesthesia

With all the possibilities brought about by AI, it’s not just radiologists who get to have all the fun. Fear not anesthetists, as there are many new and emerging technologies for us to explore and incorporate into our daily practice. These take the form of Anesthesia Information Management Systems (AIMS), pharmacological dosing systems and novel regional anaesthesia delivery systems.

Being an anesthetist requires one to record and manage large amounts of data from various input sources. AI aims to reduce this mental load by automating record-keeping and data analysis to ensure better operative conditions and patient outcomes. AI algorithms can also analyse physiological data in real-time to detect early signs of adverse events, allowing for timely interventions and enhancing patient safety. Further, they can analyse data from previous procedures to identify patterns and trends that might indicate a higher risk of complications, guiding anaesthetists to take preventative measures by highlighting high-risk patients.

The dosing and delivery of anaesthetic medications can also be improved by utilising AI systems. In fact, Target Controlled Infusion (TCI) pumps already in use for total intravenous anaesthesia actually utilise AI as complex algorithms to match flow rates based on patient metrics to maintain desired serum drug concentrations. The next evolution of this comes in the form of Closed-Loop Anesthesia Drug Administration (CLAD) systems (e.g. Diprifusor, McSleepy, SEDASYS) that use feedback mechanisms to adjust drug delivery based on real-time patient data. Research has already shown that these delivery systems based on a bispectral index (BIS) outperform manual TCI dosing in induction, achieving target anesthesia, and shortening wake-up time. Additionally, these systems could also reduce delirium during arousal. More recently, advanced multi-input multi-output (MIMO) systems, which can manage sedation, analgesia, and fluid management concurrently, are being investigated. Imagine a world in which you can simply input the desired parameters and the rest is taken care of automatically!

Lastly, by combining and integrating AI with the advancements in robotics there is the emerging possibility of automating tasks such as intubation, pre-sedation ventilation, and regional anaesthesia (e.g. Da Vinci Surgical System, Kepler Intubation System). CV technologies can also contribute to anaesthesia by providing real-time feedback and needle guidance to significantly enhance the accuracy and safety of regional anesthesia techniques in the future.

AI in Medical Research

The once mundane and dreaded tasks of literature searches, data collection, data extraction, analysis, and finally manuscript writing may soon be remnants of the past as AI technologies arise to save the day. Widely available chatbots such as OpenAI’s ChatGPT and Anthropic’s Claude AI can serve as excellent tools to assist in scientific writing. Whether it is helping improve your prose, or proof-reading , they are a godsend.

Google’s experimental new offering, NotebookLM, uses the NLP abilities of their very own Gemini Pro AI model to not only assist in writing but also streamline the literature analysis process. With it, you can upload your own sources and train your own model. Then, you can ask it questions as if it were a knowledgeable expert on the topic. This allows for time originally spent reading through pages of text, to be better spent on writing, editing and improving the article. In fact, that’s how this article was researched!

Leveraging their aforementioned capabilities, AI technologies have also been used in biomedical and pharmacological research. Researchers have been able to rapidly analyse large and complex datasets, such as genomic or clinical trial data, to identify patterns and trends that even human researchers might miss. The successful identification of new RNA-binding proteins altered in amyotrophic lateral sclerosis is one such instance. AI was also able to reliably predict the progression of Alzheimer’s disease by analyzing amyloid imaging data and accurately predicting drug therapy response.

AI in Medical Education

With younger generations being more tech-savvy than ever before, AI can be used to supplement their learning too. The previously mentioned NotebookLM once again is useful in this area as textbooks, study guides and other text-based learning materials and resources can be uploaded, quickly summarised, and searched to find answers to questions students may have. It can even create questions based on the uploaded content. It’s like Ctrl + F on steroids! AI chatbots can also promote self-study as they can serve as personal education advisors to help break down large and complex topics into more manageable chunks, thereby creating individualised study plans and roadmaps that students can then action.

AI powered platforms can also serve to make learning more engaging by generating unique scenarios and question banks adapted to individual students’ learning styles. OSCER, an Australian MedEd startup, aims to do this by utilising AI and real patient data to create virtual patient cases, allowing students to practice their history taking, diagnostic and patient management skills from the comfort of their own homes. Further, these learning systems can be extended to give personal and targeted feedback by analysing student performance. These technologies have the benefits of minimising teacher supervision, thereby decreasing costs associated with traditional teaching methods, and also creating safe learning environments where there is no risk to real patients.

The possibilities don’t just stop there. AI technologies have also been shown to be beneficial in surgical and procedural training and assessment as they can analyse data from virtual reality simulations to assess and guide trainees. The TOUCH system, for surgical training, LAHYSTOTRAIN system, which focuses on laparoscopic surgical skills, and the EchoComJ simulator, for echocardiography training, are a few examples of this in practice.

On a more systematic scale, AI can serve as curriculum review and improvement tools for large education institutions. By analysing student and staff satisfaction and performance, it can potentially identify links between lower scores and certain aspects of a curriculum, thereby informing content and structural improvements and guiding policy reform.

Limitations of AI

Despite the myriad benefits of AI that have been discussed, as measured and logical clinicians, it is paramount that we also consider the limitations and issues that come with using AI in medicine. Most importantly, there are multiple ethical questions about patient privacy, data security, algorithmic bias, and the potential displacement of healthcare professionals to consider.

Just as with any new technology, there is always a high risk of errors and inaccuracies, hence it is important that doctors properly consider the limitations and potential errors of the technology rather than becoming overly reliant on AI systems. If an error was to occur, who is to take responsibility? Is it the clinician, the developer, the institution, or the algorithm itself? As an extension of this issue, how are we to gain informed consent for procedures involving AI, and further, what changes to Goals of Care decision-making processes and policies need to be implemented if AI technologies were to become widespread in clinical medicine? Lastly, what is the impact of adopting these technologies on the patient – doctor relationship? Can the trust, empathy and human connection that patients have come to expect be maintained?

Due to the inherent nature of how AI models are developed, there is an innate risk of error and bias from the quality of data used to train the model. Ironically, this makes it more human in a way, as it is similar to how an individual may make mistakes if they had been taught or learnt incorrect material. This was an issue noticed in the creation of this article as Google’s NotebookLM, which was used for the research, could be biased by the sources that were selected and uploaded. Hence, it is imperative that these models developed to be used in healthcare are trained in a systemic and standardised way using data that is evidence-based and widely accepted.

Message from Google’s NotebookLM disclosing possible risk of bias and inaccuracies.

Given how novel AI is currently, its successful implementation will require overcoming technical, logistical, and cultural barriers within the healthcare sector. Currently, better infrastructure such as improved data storage, computing power and energy sources to power these systems is required. The seamless integration of AI systems into existing clinical workflows and electronic health record (EHR) systems is also essential for widespread adoption as it will reduce the barrier to entry and make AI more accessible to clinicians. Moreover, as AI systems rely on access to large amounts of personal and sensitive data, such as medical records, outdated cybersecurity safeguards need to be scrutinised and improved upon to protect and respect patients’ confidentiality and to avoid identity theft or fraud. With many cybersecurity breaches plaguing Australians in recent times, this is a major issue that needs to be addressed. Lastly, the high cost associated with developing and implementing these AI systems may be a limiting factor as it can deter healthcare providers and institutions from adopting them into regular practice. For patients, there is also a risk that these costs may perpetuate and exacerbate pre-existing disparities in access to quality healthcare.

Similarly, again due to the novel nature of AI, educational curricula need to be updated to include information and guidance for medical students on understanding how AI models work and how to use these models. Perhaps, in the future, as AI integration becomes more mainstream, there may also need to be education on decision-making about when to use these technologies due to the dangers associated with their over-reliance and over-use.

Lastly, as medical education is far removed from the fields of information technology and artificial intelligence, doctors are often unaware of how AI systems work. These systems essentially operate as “black boxes” as their decision-making processes can be opaque and difficult for clinicians to interpret or understand. This lack of transparency can make it challenging to identify errors, build trust in the system’s recommendations, and explain the rationale behind a particular decision to patients. Hence, it is paramount that if AI is to be implemented in healthcare, it needs to be transparent and easily understandable, at least in the short term, as medical education catches up to fill this gap in clinical knowledge.

Final Verdict

Just as the Internet revolutionised the way we interact with our world, AI may act as the new messiah of our future. With it, there is limitless potential for development and progress, expediting our evolution as we as a species challenge and explore the mysteries of our universe. In the medical realm, AI has the ability to not only improve patient care and outcomes but also guide the next generation of clinicians and researchers to achieve greatness. Yet, despite its many benefits, it is imperative that we take a measured approach to its implementation and adoption as the ethical and systemic risks and limitations associated with AI technologies are yet to be fully understood, comprehended and addressed. We look forward to the future that awaits us- one in which we will work alongside AI to make our world a safer and more equitable place.

References

Antel R, Sahlas E, Gore G, Ingelmo P. Use of artificial intelligence in paediatric anaesthesia: a systematic review. BJA Open [Internet]. 2023 Feb 7 [cited 2024 Jun 19];5:100125. Available from: https://www.bjaopen.org/article/S2772-6096(23)00003-5/fulltext

Bowness J, El‐Boghdadly K, Burckett‐St Laurent D. Artificial intelligence for image interpretation in ultrasound‐guided regional anaesthesia. Anaesthesia [Internet]. 2020 Jul 29;76(5):602–7. Available from: https://associationofanaesthetists-publications.onlinelibrary.wiley.com/doi/10.1111/anae.15212

Chan KS, Zary N. Applications and Challenges of Implementing Artificial Intelligence in Medical Education: Integrative Review. JMIR Medical Education [Internet]. 2019 Jun 14;5(1)
. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6598417/

Civaner MM, Uncu Y, Bulut F, Chalil EG, Tatli A. Artificial intelligence in medical education: a cross-sectional needs assessment. BMC Medical Education [Internet]. 2022 Nov 9;22(1). Available from: https://link.springer.com/article/10.1186/s12909-022-03852-3

Gupta B, Sahay N, Vinod K, Sandhu K, Basireddy H, Reddy Mudiganti R. Recent advances in system management, decision support systems, artificial intelligence and computing in anaesthesia. Indian Journal of Anaesthesia [Internet]. 2023 Jan 21 [cited 2023 Jun 18];67(1):146. Available from: https://journals.lww.com/ijaweb/fulltext/2023/01000/Recent_advances_in_system_management,_decision.22.aspx

Han ER, Yeo S, Kim MJ, Lee YH, Park KH, Roh H. Medical education trends for future physicians in the era of advanced technology and artificial intelligence: an integrative review. BMC Medical Education [Internet]. 2019 Dec 11;19(1). Available from: https://bmcmededuc.biomedcentral.com/articles/10.1186/s12909-019-1891-5

Holzinger A, Langs G, Denk H, Zatloukal K, Müller H. Causability and explainability of artificial intelligence in medicine. WIREs Data Mining and Knowledge Discovery [Internet]. 2019 Apr 2;9(4). Available from: https://wires.onlinelibrary.wiley.com/doi/full/10.1002/widm.1312

Kaul V, Enslin S, Gross SA. History of artificial intelligence in medicine. Gastrointestinal Endoscopy [Internet]. 2020 Oct 1;92(4):807–12. Available from: https://www.giejournal.org/article/S0016-5107(20)34466-7/fulltext

King MR. The Future of AI in Medicine: A Perspective from a Chatbot. Annals of Biomedical Engineering [Internet]. 2022 Dec 26;51(2). Available from: https://link.springer.com/article/10.1007/s10439-022-03121-w

Masters K. Artificial intelligence in medical education. Medical Teacher [Internet]. 2019 Apr 21;41(9):976–80. Available from: https://www.tandfonline.com/doi/full/10.1080/0142159X.2019.1595557

McKendrick M, Yang S, McLeod GA. The use of artificial intelligence and robotics in regional anaesthesia. Anaesthesia [Internet]. 2021 Jan 10;76(S1):171–81. Available from: https://associationofanaesthetists-publications.onlinelibrary.wiley.com/doi/full/10.1111/anae.15274

Mohammad Muzaffar Mir, Mir G, Nadeem Tufail Raina, Mir S, Mir S, Elhadi Miskeen, et al. Application of Artificial Intelligence in Medical Education: Current Scenario and Future Perspectives. PubMed [Internet]. 2023 Jul 1;11(3):133–40. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10352669/

Naaz S, Asghar A. Artificial intelligence, nano-technology and genomic medicine: The future of anaesthesia. Journal of Anaesthesiology Clinical Pharmacology [Internet]. 2022 Jan 28;38(1):11. Available from: https://journals.lww.com/joacp/fulltext/2022/01000/artificial_intelligence,_nano_technology_and.3.aspx

Ni L, Lu C, Liu N, Liu J. MANDY: Towards a Smart Primary Care Chatbot Application. Communications in Computer and Information Science [Internet]. 2017 Oct 17;38–52. Available from: https://link.springer.com/chapter/10.1007/978-981-10-6989-5_4

Rampton V, Mittelman M, Goldhahn J. Implications of artificial intelligence for medical education. The Lancet Digital Health [Internet]. 2020 Mar;2(3)
–2. Available from: https://www.thelancet.com/journals/landig/article/PIIS2589-7500(20)30023-6/fulltext

Singh M, Nath G. Artificial Intelligence and Anesthesia: A Narrative Review. Saudi Journal of Anaesthesia [Internet]. 2022;16(1):86. Available from: https://journals.lww.com/sjan/fulltext/2022/16010/artificial_intelligence_and_anesthesia__a.16.aspx

Singhal M, Gupta L, Hirani K. A Comprehensive Analysis and Review of Artificial Intelligence in Anaesthesia. Cureus [Internet]. 2023 Sep 11;15(9). Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10566398/

Winkler-Schwartz A, Bissonnette V, Mirchi N, Ponnudurai N, Yilmaz R, Ledwos N, et al. Artificial Intelligence in Medical Education: Best Practices Using Machine Learning to Assess Surgical Expertise in Virtual Reality Simulation. Journal of Surgical Education [Internet]. 2019 Jun 13;76(6):1681–90. Available from: https://www.sciencedirect.com/science/article/pii/S1931720419301060

It seems that regional anaesthesia is somewhat past its heyday. 

Many hospitals have adopted other techniques for pain management in operations that were once common practice grounds for regional, hospitals don’t have the systems set up to facilitate fast turnover and having sufficient time for regional anaesthesia onset and there are no studies to prove convincingly any long term morbidity or mortality benefit. 

However in my experience this is one of the most incredibly useful skills in your anaesthesia quiver!

There will be many times in your career that you will be required to anaesthetise an incredibly sick patient for a peripheral/limb procedure. 

Imagine a patient from ICU requiring an urgent open reduction and fixation of his fractured ankle. He is 150kg, on BIPAP for pneumonia, heart failure with EF30% and saturating at 90% due to his severe COPD. Performing a general anaesthetic for this operation is definitely an option and will require a lot of prep time and may result in airway management difficulty, CVS instability and oxygenation risks. And this situation isn’t that uncommon! It will arise time and time again whether it’s an upper limb procedure, lower limb procedure or the elective arterio-venous fistula in a sick renal patient. 

I would argue your ability to perform effective regional anaesthesia improves welfare for the patient and ease of providing effective anaesthesia for you and your team!

So with fewer opportunities to provide regional anaesthesia how do you gain enough experience? If practice makes perfect how do you get enough practice?

It is vital to manufacture practice and create opportunities!

What do I mean? Well there are a few select skills with ultrasound (US) guided regional anaesthesia that you can master even with minimal practice at actual regional anaesthesia. ‘Chunking’ these skills and practising them separately is one of the best ways to learn difficult and uncommon techniques

These skills are:

1. US scanner competence
2. Sono-anatomy mastery
3. US-Needle coordination

US scanner competence

• Bring out the US scanner at every opportunity to scan nerves, veins, arteries and hearts
• Learn a systematic way of improving your image. 
  • Select the correct probe
  • Select the depth
  • Optimise the gain (how much signal or ‘whiteness’ comes back)
  • Select the focal point
  • Probe manipulation: practice sliding, tilting, rotating and translating

Sono-anatomy mastery

Sono-anatomy is like exploring a new neighbourhood. The first few times everything is unfamiliar and you won’t recognise much. You then decide to buy a map and explore your new neighbourhood every day and suddenly you will start to recognise signs, roads, laneways and your favourite cafes and shops. 

The way I increased my familiarity with the sonoanatomy neighbourhood was to scan a patient every day. My map was Ultrasound Imaging for Regional Anaesthesia from www.usra.ca/education/booklet.php (electronic copy available on iTunes). This is an exceptional resource, which shows the labelled cadaveric and sono-anatomy so you can trace the nerves from proximal to distal in the upper and lower limb. 

Put a probe on the consented patient and try the following. For your lower body cases try tracing the brachial plexus from the interscalene, to supraclavicular, infraclavicular and axillary views and then trace the terminal nerves (ulnar, median and radial) from axilla to the wrist. Likewise during your upper body cases you can ‘walk’ around the lower limb sono-anatomy neighbourhood tracing the femoral nerve and sciatic nerves. Use Ultrasound Imaging for Regional Anaesthesia to show you the relevant structures.  

After doing this a few times you will gain familiarity of the  

• nerve plexuses
• nearby structures, 
• the way nerves often follow vascular structures and
• the way they differ from tendons and muscles.
• With even more experience you will notice the inter-patient variability and abnormal anatomy. This becomes increasingly relevant when you notice that some patients are difficult to block because they have aberrant nerves, tissue/septae that divide plexuses or vascular structures running nearby.

Another great way to have guided practice is to sign up to one of the many US regional anesthesia short courses. 

Other useful resources include:

• www.nysora.com
• www.asra.com
• AnSo app

US needle coordination

There are many techniques that are common in anaesthesia and enable us to practice US needle coordination. After you have mastered cannulation try performing cannulation with US guidance. This is a great way to improve your success with difficult cannulation but also a great way to learn how to coordinate your needle manipulation in one hand and the US probe in the other. 

Other slightly advanced techniques to improve US coordination are US guided arterial line insertion and central line insertion.

In my experience there were some relatively common opportunities for US guided regional anaesthesia. The fascia iliaca blockand femoral nerve blockare considered relatively safe and easy techniques and where I developed most of my skills in US coordination. They are superficial, the landmarks are easy to find, have few critical structues and inadvertent vascular puncture can be compressed to stop a haematoma. When clinically indicated, do these regional blocks with US guidance.

So there is it! My method for trying to increase your learning curve, when your learning opportunities might be few and far between.

Any questions or comments please email me 

Anaesthesiacollective@gmail.com

https://riskcalculator.facs.org is an exceptional resource to provide reasonably accurate and patient specific risk information to help guide your perioperative decision making.

This calculator uses a series of 20 questions and the surgical procedure to provide 18 different outcomes within 30 days of the procedure. These include risk of death, serious complications and even the chance of whether the patient will need postoperative rehab.

Before this calculator existed I felt that I had a lack of insight into the risks outside of the intra-operative anaesthesia risk. As anaesthetists, we rarely follow up our patients and may not be aware of the patient’s risk a month after the operation. Even our surgical colleagues may not have great data on their patient’s risk profile on such a large scale.

This calculator was built using data from over ‘4.3 million operations in over 730 participating hospitals from 2013-2017‘

I find the calculator most useful when I have a particularly unwell patient. I am able to quickly plug in the requested data points and then assess the risk. Have a look at the risk profile below for this imaginary elderly male undergoing a laparoscopic cholecystectomy.

I find this incredibly useful information to guide my perioperative management and providing informed consent to the patient.

For example I may

• refer to HDU/ICU preop
• discuss risk factor modification and use it as a ‘teachable moment’
• use the data to highlight how high risk (or low risk) something might be to a patient or colleague who has a different opinion on the risk level
• plan post op nursing or rehab requirements
• choose to recommend a particular operation in a centre with ICU/cardiology support
• discuss the risks and benefits with more evidence than limited by my own experience

any questions please comment below!

I was recently supervising an exceptional trainee. He did everything right to make the most from our operative list together. He texted me the night before with things he wanted to learn, He arrived early to the list and had everything checked and set up. I was really impressed with his motivation and assertiveness and knew he would have a rapid learning curve. Now as it happened he also missed an intravenous cannula and wasn’t able to insert the spinal needle. What was really interesting was that this was quite upsetting for my trainee but my impression was that he was very competent and reliable doctor in training.

My assessment of his ability and his assessment of his ability were worlds apart!

I now think back to my own training, and how important it seemed that we are always ‘successful’. We are praised for success and often berated if we don’t succeed, even though many of our technical tasks as anaesthetists can be very humbling.

While we do improve over time, there will be some veins that we just fail to access, arterial lines that never seem to thread and spinals where we just aren’t able to find a route into that elusive intrathecal space.

But what I suggested to my trainee is that the path to success isn’t repeated success…

The path to success is a diligent process of repeated failure

I could now write a whole bunch of cliches or post some inspirational memes proving my point but you’ve probably already seen these, but one does stick to mind… you would never think an infant failed as they stumbled trying to walk. It is simply the process that gets us to a point of competence.

I hope that most of us (trainees and supervisors) realise the critical important of rewarding a diligent process and showing very little if any frustration or negativity if the initial outcome isn’t success.

I believe that this would foster a more productive culture for learning, development and welfare.