We like to start our work at a project with a top down analysis to find out if and how we can be of help. As a partner to you we thrive on the success of your organization. These steps help us understanding how to best serve you.
Is now the right time to look into Artificial Intelligence solutions for my organization?
With few exceptions the answer to this question is yes. Across all industries digitalization and implementation of artificial intelligence solutions are becoming competitive advantages. Falling behind in setting up effective critical business processes is increasingly risky. Seven years ago artificial intelligence solutions in healthcare was very rare, bordering unheard of. Three years ago still unusual. Today, most organizations we talk to have either implemented some predictive analytics solutions or a in the middle of a transformation. What was an edge yesterday will be a necessity tomorrow.
Define what problem Artificial Intelligence is helping to solve?
Healthcare and medicine are riddled with many problems that with the right approach turn into data science opportunities. Here are a few examples.
- Cost of care
- Inability to meet patient expectations
- Staff burnout
- Increasing demands for documentation
- Complex and time consuming workflows
- Compund development and affordability
- Customer retention and cost of acquisition
Analyzing and anchoring the key issues of an organization internally is time well spent. Once this step is taken it makes perfect sense to start looking into artificial intelligence as a potential solution. Solutions are typically on a case-by-case basis but in general Artificial Intelligence is best applied to large-scale problems involving quantities of patients, staff or routines. We often help free up time for patient interactions.
Analyze the true costs of the problem?
Putting a dollar value to the cost of the problem is a very good exercise. In the end of the day all organizations need to justify an investment by a return on interest. Accenture has reported in some industries, AI investments are set to boost revenues by over 30% over the next four years.
If possible have a controller or an accountant help you look into the dollar value. It is easy to miss indirect costs such as recruitment costs, capital costs on machines etc. As humans we seem to be prone to underestimate time spent on various tasks.
With a decent ballpark estimate of the cost of the problem you have all you need to move ahead with the decision to look into an artificial intelligence solution or not. Next step precision. Everything in artificial intelligence boils down to precision. How good is the technology at performing the task? A useful indication is if it can match a human. Then the cost for salary and payroll expenses for that resource utilization is the value. Either way a threshold for the precision and performance of the artificial intelligence solution is critical. An open-ended pilot or general feasibility study is often destined for the drawer.
State how AI can be used by your organization?
Perhaps you are already acquainted with the technology and have a few applications running or you are just starting up the process. There are a couple of action points to notice.
Start looking at data in the context of areas of pain. If you already have abundant data and preferably machine-collected data in an area of pain this is a good start. Next have a look at your workflows and talk to the organization that work in this particular area. Do they see a potential for a change? Invariably the result of introducing artificial intelligence solutions leads to a reengineering of business processes and a significant change. Downsizing can be one outcome but more common is an automation of mundane tasks and an improved work situation. A key to a successful transformation and an ROI on investment is to get the team and other stakeholders onboard this process day one. Program management comes into play and often these discussions involve C-suite executives as well as the board of directors. Artificial Intelligence is not a plant wall.
When there is commitment it is time to dedicate staff to help build the artificial intelligence solutions. Someone with clout and access to decision makers facilitates change and someone accountable at the IT department is also critical.
Avoid getting lost in details or caught up in the maze of causality
Working with artificial intelligence brings pitfalls. One common pitfall is getting lost in details. Few people have a full comprehension of the math and logics behind artificial intelligence and spend an unnecessary amount of time getting into the nitty gritty. In such a fast paced environment as data science it is difficult for outsiders to delve deep enough into the technology and keep up to date with the changes. Put your focus on the needs of your organization and how the business process reengineering will make you more productive and you will be headed for success.
Another trap is the causality trap. Trying to understand every single step in value creation or putting an exact and constant value to the technology is often a tremendous time trap; especially so in healthcare. Data such as claims change over time and the truth this week will certainly be different in three or nine week’s time. When will we truly know? Breaking down value creation between humans and artificial intelligence in clinical decision support systems can lead to a philosophical challenge in establishing scientific theory.
Carefully select the key KPIs resonating with your organization and have faith they will guide you through the process. Areas of interest include honest time management assessment and top dollar count such as average dollar claim per patient, number of emissions, number of primary care visits etc.
Implement the AI solution
Obviously beating our own drum we recommend partnering with an AI consultancy company for the implementation. AI implementations are so complex by themselves and oftentimes pilot projects create domain specific challenges to solve along the way. In addition there is the process change and the inherent organizational change that goes with it.
Going with a vendor that has made software installations including AI before will both bring most experience and seniority to your project and affect outcome. The number of data scientists has grown very quickly: there are now far too many data scientists with very little experience heading into a job market that has very few experts, leaving us with a ‘bottom-heavy’ candidate pool. Facing this reality ascertain you have an experienced team building your solution. This allows your team to focus on the roll out and monetizing the change while you can hold the seasoned vendor accountable to the timeline and performance.
Encourage your change transformation team and celebrate wins
Business process reengineering can be both risky and difficult. There is the political risk by sticking one’s neck out and getting tied to a project that may not be favored or by default as successful as intended. Bear in mind, also AI projects carry an implementation risk. Before this wonderful AI summer of deep learning there have been numerous AI winters. At times AI professionals even shied away from being labeled data scientists.
So make sure to adhere to fast prototyping and include reality checks. Prep your team and equip them to manage difficulties and empower them with relevant KPIs. Above all encourage them publicly for their work. New rules zero in on algorithmic powered business with cost cutting or data driven revenue generation. Teams that can pull this off are worth its weight in gold so reward them generously and celebrate wins.