By Mark Watts
We are at the beginning of the transformation for medical imaging artificial intelligence (MIAI). Currently, less than 30 percent of U.S. health care providers are using it. I am concerned about training the staff to support this new innovation. Will the current PACS administrators be tasked with the support and care of MIAI?
It is true most hospitals are still struggling to find employees with the skills necessary to create, train and work alongside intelligent machines.
As health care practices become aware of the efficiency gains that can be achieved through leveraging the power of machine learning, computer vision and similar technologies, demand for skilled workers in the field is quickly outstripping supply.
The American College of Radiology, RSNA and universities have responded to this by creating new courses and educational programs focused on these skills. But anyone wanting to break into the industry may still be confused at the options available to them. So, here’s a rundown of some of the most valuable skills you can learn today if you want to be prepared to work with the automated, intelligent machines of the future!
Visualization and Skills Communication
It’s great to be able to use computers to make decisions and attain a deeper understanding of complicated subjects than would ever be possible using purely human-scale analytics. However, if we don’t have the ability to communicate those findings to other humans – and explain why they are so valuable – then it’s all a waste of time. Many organizations have proven that it’s possible to bring about widescale, positive change – both internally and across societies as a whole – by utilizing AI and machine-driven decision-making. But communication skills are essential to generating the buy-in necessary to reap the benefits. This is the reason that “data communicators” and “data translators” are one of the most in-demand sets of skills when it comes to AI and machine learning in business right now. Strong visualization skills mean the ability to take the insights uncovered by machine learning tools and convert them into compelling storytelling that communicates exactly what needs to be done, when, and by whom, in order to achieve growth and results
This is a relatively new term that has emerged in recent years to cover the skills needed when it comes to working with the plethora of AI-related tools and services that have become available. AI Ops involves administering and managing all of the connected systems that go into delivering modern AI infrastructure, in order to ensure continuous uptime and a good level of service to the end-user, which could be the business itself or its customers. It might involve coordinating the use of a number of AI-as-a-service elements that connect together to create the organization’s AI infrastructure. AI Ops also refers to the process of administering or overseeing AI analytics of an organization’s IT and data operations. This could involve implementing machine learning processes to enable more efficient use of data within the organization or its IT infrastructure as a whole.
Data is absolutely fundamental to the ability of machines to think and learn. Data is the input used to train AIs to make decisions and carry out tasks. Data scientists understand how to capture, manipulate and work with data in order to extract insights from it. These skills are essential to the field of AI because they encompass the advanced analytics that are necessary in machine learning algorithms. Data science has been a part of computer science educational curriculums for a long time, and today they are usually heavily focused on applying AI to solving business problems using available information.
These are old-fashioned mathematical skills that are still considered essential for anyone who wants to understand how AI works, why it is useful, and where it can be most usefully deployed. Techniques such as linear regression, logistic regression, clustering, Bayesian modeling and random forest analysis were all around long before AI became a buzzword. AI performs the core task of making predictions based on identifying patterns and spotting outliers. Probability and statistics are still at the heart of many of the most sophisticated AI algorithms. Understanding the principles behind how they work is key to understanding why computers are such powerful tools when it comes to automating decision-making. A firm understanding of statistics and probability is hugely valuable when starting out in AI. It helps us to understand how to articulate problems and propose solutions by selecting the most appropriate models and techniques.
Although no-code and low-code AI solutions are appearing that let us leverage AI solutions without getting our hands dirty, it’s likely that businesses that want to deploy their own bespoke AI solutions will require skilled coders for a long time yet. A basic understanding of at least one of the most popular programming languages for AI – Python, R, C++, and Java – is very useful for anyone working with machine learning algorithms. This may seem a little counter-intuitive because the purpose of AI is to enable computers to “learn” without having to be specifically coded to carry out a job. Nevertheless, most people working in roles that involve AI today recommend some level of experience in coding for anyone wanting to prepare themselves for using AI.
If you need guidance as to what courses I would suggest or what courses from MIT, Stanford or Royal Academy that I have taken I am at your service. •
Mark Watts is experienced in the imaging realm and is the founder of Zenlike.ai.