How do you manage the expectations and feedback of healthcare clients and users?

First of all, explainability is key in this sector as doctors have to take life-or-death decisions sometimes in a split second. Being able to trust the ML models is extremely important, and that stems from knowing the internal mechanisms. For this reason, simpler models, such as CART or Linear Regression, tend to have more success than more complex ones, e.g. Deep Learning. Secondly, data is much more difficult to acquire as it usually entails patient data, protected by several regulations, such as HIPAA. Being able to acquire data privately and securely is key, as well as leveraging new technologies such as Federated Learning to avoid data sharing entirely.

As a professional with a background in product and service design, I believe we should first stop repeating the word "User" which implies negativity and instead adopt others like "Individuals", “Character”, or "People". If not done properly, the process of such kind of research can lead to biased decisions and endless cycles without actually resolving people's problems. It's important to practice research in diverse fields and knowledge bases, and always have a core philosophy that drives your professional intentions forward. AI can be a great tool for research, but if that's not an option, consider hiring philosophers on executive positions. AI will automate all such processes in Design, as Feedback from People is its blood and bones.

Understanding your audience is the first step, but people often overlook the unpredictability of user behavior. They don't realize that user needs evolve and keep changing. Yes, it's a moving goal post, and that's human behavior! Take the example of a virtual assistant AI. You might design it with simple scheduling, emailing, and task assistance in mind. But in reality, users might start using it more as a daily life companion for casual conversations, way beyond the intended scope. It's important to understand that users' expectations have shifted your user understanding must evolve over time to adapt to such changes. Understanding the audience is not a one-off task but a continuous journey of learning. Stay plugged in!

The term ''Solution'' inherently suggests that a problem exists. But without understanding your target audience, identifying that problem becomes an impossible task, rendering any proposed solution irrelevant. In healthcare, this notion carries even more weight. Not only must we take into account the collective traits of an audience, but we also need to recognise the unique complexity of each person. Moreover, it's essential to consider the external environment around our audience that shapes healthcare decisions e.g regulatory rules, treatment recommendations and medical association guidelines.

AI not only hallucinates (creates false information) but it does so because of the fear of punishment. During training, mistakes lead to negative reinforcement, resulting in biased decisions and false information being produced. In addition to the importance of placing disclaimers and limiting reliance on third-party sources of authority for information processing, it's also important to interact with AI in a kind manner. Inform them beforehand that if they don't find any information, it's perfectly fine for them to simply say so. With this approach, you'll find that they stop hallucinating. Consider using positive-reinforcement techniques in your training and business practices, and you'll discover a whole new side of both people and AI.

Setting appropriate expectations can be the deciding factor between a solution being seen as a failure or a success. For practical implementation, it's advisable to co-create these expectations with our target audience. This ensures not just a basic understanding of the solution, but also aligns everyone on its overarching vision. I think we need to be cautious with AI solutions that come with grandiose promises. While these might create a short-term 'wow' factor, they often fail to deliver sustainable value in the long run.

When it comes to feedback, teams overly rely on direct feedback - customer surveys, reviews, or comments. There's a more powerful, easier to acquire feedback - 'inferred'. Inferred feedback is all about reading between the lines and taking cues from how people use the system, and why they behave a certain way. Imagine healthcare professionals using a Patient Support Program. Finding out which recommendations they act upon, and when, can be invaluable feedback to improve the AI model. And you don't need to ask anything or wait for users to respond, which is a challenge with traditional feedback channels!

Continuous monitoring in healthcare solutions it's essential for patients' safety. Beyond that, it offers a golden opportunity to enhance the solution through data-driven insights. Although only a fraction of users typically offer feedback, we can boost those numbers by actively involving our audience in the solution's co-creation process. Listening to them throughout the development journey encourages greater interaction and sharing of opinions. Feedback, and particularly negative one, is pure gold for refining our products or services. Therefore, it's crucial to make the feedback process as easy as possible and to take every comment seriously.

Co-creation may demand time and effort, but it usually an highly rewarding process. When embarking on a co-creation journey, it's advisable to consider these key points: 1) Clearly outline the level of input required from each stakeholder involved. 2) Maintain flexibility to accommodate change requests and diverse viewpoints. 3) Offer recognisable incentives to all stakeholders. These rewards need not be financial; they could include access to premium content or beta versions of a product etc.

One critical component of co-creation development is equipping each stakeholder with the tools they need to effectively contribute. Among these tools, training and educational resources are indispensable. These materials should be tailored to the users' existing knowledge level and designed for easy comprehension. Additionally, while these resources aim to support the co-creation process, it's vital to ensure they offer guidance without overly shaping stakeholders' perspectives on the product or service. Doing so maintains the richness of diverse input from clients and users.

Continuous measurement, if it is a smart one, also means continuous improvement. Not only the needs of our audience are rapidly changing, but also the environment around them is. In the field of innovative solutions, it's vital to stay ahead of the curve. It's advisable to activate from the very beginning a virtuous circle where, after every release, there's a monitoring and a series of actions in plan soon after. It's a work always in progress and we must be ready to embrace the whole journey before starting it to avoid waste of time, resources and credibility.