There is no doubt that the AI race is officially on in healthcare. The sector has long had a reputation for lagging behind other industries when it comes to embracing digital transformation. However, 2024 is set to be a major turning point, with many healthcare companies committing to increasing their budget on AI.
A recent survey among providers confirmed that AI spending is up by as much as 80%, with almost half (43%) of respondents investing more than $300,000 in the last 12 months alone.
However, many organizations may even feel they have little choice in instigating change to keep up with demands from patients, payers, and regulators who seek improved care, safety, and access.
So, with many companies about to invest significant resources into their AI journey, let’s take a look at some of the pitfalls to avoid.
1. Investing in generative AI: Generative AI such as ChatGPT has made huge waves in the democratization of AI, with everyone from the C-suite to frontline employees and students being able to use it through plain-language commands. This unprecedented agility and accessibility to citizen developers helps explain why 94% of upper management believe that these large language models (LLMs) like ChatGPT will be transformative to their business. In healthcare, however, investing in generative AI and LLMs cannot happen on a whim.
Regulatory scrutiny is likely to intensify to ensure the safe and ethical use of AI in healthcare. This could encompass rigorous validation of AI solutions and large language models to ensure accuracy, transparency in AI decision-making, and adherence to patient data privacy laws. This increased scrutiny will not only come from regulatory bodies but from patients themselves who may not trust it. When it comes to compliance, both in the EU and the US, smaller specialized AI models fall into different categories of these regulations.
Therefore, healthcare leaders should make use of specialized platforms that can adequately fulfill their needs without the additional risk associated with LLMs. Purpose-built AI solutions, often referred to as narrow AI, can be developed to address specific medical challenges such as disease diagnosis, treatment planning, and patient management. Unlike general AI, these purpose-built solutions are tailored to adhere to medical regulations and compliance standards and ensure patient safety, making them more suitable for healthcare applications. Narrow AI models that can extract, classify, and automate patient-outcome information into EHR systems are already available, eliminating the need for a more costly, cumbersome, and risk-prone generative AI.
2. Failing to train staff properly: Research shows employees’ lack of training is a leading cause of digital transformation not being successful. While frontline workers represent over 70% of the U.S. workforce, only 14% say they have received training on how AI will affect their jobs, according to a new study. It’s therefore imperative that employees get as much support as possible. The growth of low-code/no-code AI tools now enables many healthcare leaders to make digital transformation possible without spending additional resources or relying exclusively on IT specialists.
Healthcare IT leaders must ensure that IT vendors providing software solutions offer the right skills training, not only during the transformation stage but also with ongoing support post-implementation. This may include practical workshops or training sessions where employees can interact with the AI tools directly. Hands-on experience helps them better understand the functionalities and nuances of these tools in real-life business scenarios. You can also provide customized training modules tailored to different departments or specific technologies such as machine learning and provide ongoing access to resources to keep employees up to date with evolving trends. There are also open-source tools available as well as more intense offerings from the likes of Coursera, Udemy, and edX covering relevant topics such as machine learning, deep learning, and AI applications.
According to a Gartner report, 70 percent of new apps built will use low-code or no-code technologies by 2025. For business workers in healthcare, it will mean becoming ‘citizen developers’ to create and build suitable apps that use ‘drag and drop’ technology. For example, they will be able to design, train, and digitize a wide variety of documents such as hand-printed registration forms, insurance information, or patient signatures for automatic integration into the EHR workflow.
3. Automating the wrong processes or broken processes: The need for a data-driven analysis before launching a digital transformation project is paramount, especially given the results of research by McKinsey showing as much as 70% of automation projects fail. Studies confirm one of the biggest reasons behind failure was ‘vague goals,’ with 7% even confessing to automating the wrong technology altogether. Unfortunately, IT leaders often get caught up in the noise and hype of new tools and the functionalities that can be added, pulling them away from actionable business goals and solvable real-world challenges.
To assess where you want to go, you first need to look at where you’re at to uncover where best to inject your resources. To achieve this, companies need process intelligence to gather insights and inform strategic improvements, rather than relying on feedback from management and staff who may have biased and inaccurate perspectives. By seeing your workflow as a digital twin in real time and discovering how your processes, people, and technology come together, you will be able to understand what’s working well and what must be improved to address pain points and devise a happy path to optimization.
A great example of this is referral management. I found it incredible to hear that just 54% of faxed referrals in healthcare result in a scheduled appointment, meaning the loss of millions of dollars of income and potential adverse impacts on patient health. What happens to the other 46% of patients who don’t get a scheduled appointment? One of the biggest problems is that hospitals and physicians are unaware of exactly how the process works, with manual and repetitive tasks causing costly mistakes and delays. A recent Chime-Cerner survey indicated nearly 40% of provider participants are losing at least 10% of patient revenue to referral leakage. Non-processed referrals cost hospitals between $821k to $971k per physician per year – all because of outdated manual processes.
Ultimately, when beginning an AI journey, healthcare IT leaders must remember not to get caught up in the noise of great tech demos with amazing capabilities, and instead look for a trusted advisor who will help you solve specific business challenges and improve life for both employees and patients.
About Maxime Vermeir
Maxime Vermeir is Senior Director of AI Strategy at intelligent automation company ABBYY. With a decade of experience in product and technology, Maxime is passionate about driving higher customer value with emerging technologies across various industries including healthcare. His expertise from the forefront of artificial intelligence enables powerful business solutions and transformation initiatives through large language models (LLMs) and other advanced applications of AI. Maxime is a trusted advisor and thought leader in his field. His mission is to help customers and partners achieve their digital transformation goals and unlock new opportunities with AI.