Takeaways
Decoding alphabet soup
- With so many applications, and numerous models to be built upon, it can be challenging to understand Artificial Intelligence (AI) and where it fits within the care delivery landscape.
- AI is a broad umbrella describing technology’s ability to execute human behavior.
- Machine learning is a type of AI in which technology that learns from data sets without the need for continual programming. Machine learning identifies trends and correlations in data, using these as benchmarks to set its course of learning.
- It’s common for AI to get compared to sophisticated excel sheet algorithms. The difference however, is that the excel sheet doesn’t learn from the data, but instead executes the exact functions its instructed to do. AI, on the other hand, takes data and executes functions based on trends in the data.
Today's opportunities to improve care delivery with AI
- As humans, we’re inherently biased. We filter the world through a lens that reflects our unique experiences. AI helps to remove biases from data analytics by making decisions based purely on data rather than human assumptions.
- When it comes to diagnostics and preventative outreach, this objective analysis is crucial because it expands our understanding of who is at risk for which diseases.
- In cardiology, for example, we commonly assume that those over the age of 55 have an increased risk of heart attack. AI could disprove or build upon this assumption, to include other at-risk populations.
- Chatbots also present an opportunity for health systems to improve care delivery because they offer an accessible way for patients to interact with their health system. With the right authentication layers and connectivity to other data sources, AI could address the user’s intent and channel it in the right way without burdening staff.
AI advancements to keep an eye on
- While Elon Musk may have predicted 2025 as the “year of singularity,” Chris Hemphill, VP Applied AI & Growth at SymphonyRM says it’s too soon to tell because unknowns will undoubtedly appear.
- One direction AI could take us – a direction we’re just on the fringes of now – is proving direct causation, being able to definitively identify why certain things are happening.
- Domain experts and clinicians will need to be increasingly involved in AI application to make sure learning models are founded on a true understanding of the practice and context, utilizing the right questions and labels.
- There will likely be an increased focus on weeding out biases from data collection. If the data that AI analyses reflects biases, it will embed those biases in all of its outputs. In the future we will see increased focus on collecting unbiased data.
