The current problem with AI in Health Tech: great applications, disconnected experiences

Foto del autor

By Damian Calderon

May 28, 2024

The latest AI revolution, which includes Generative AI and Large Language Models (LLMs) presents an extraordinary opportunity to address bottlenecks in health tech. These technologies offer the potential to personalize and summarize information, streamline operations, and translate between natural language and technical terms. But a smooth integration needs cross-collaboration and connection with practices and patients’ realities.

Overview of AI Applications in Health Tech

AI has made significant strides in health tech and is on its way to transforming areas such as diagnostics, treatment personalization, and operational efficiency. For instance, by automating routine tasks like writing chart notes, AI allows clinicians to focus more on patient care, improving job satisfaction and reducing burnout. However, these advancements often remain isolated, failing to create a cohesive and integrated experience for healthcare providers and patients. This isolation has far-reaching effects on various stakeholders. Patients may face fragmented care and confusion from inconsistent information. Healthcare providers might face increased workloads as they struggle to bridge the gap between AI tools and traditional workflows. Administrators could see reduced efficiency and missed opportunities for improved patient outcomes and operational savings. In a nutshell, this disconnection can lead to fragmented care applications, which is already a huge problem in healthcare.

Root Causes of Disconnected Experiences

The main reason for disconnected experiences is the lack of a user-centered design approach. ML Teams focused solely on generating successful outputs from their models will create tools that don’t align well with the practical needs of healthcare providers: a prediction alone isn’t sufficient; your AI solution must be tailored to the actual usage context, which often includes multiple layers of variation and complexity. Knowing the tools they are currently using, deciding how the generated predictions should integrate with them, and understanding more of the work context are crucial to better connecting the experience.

Also, many companies aiming to add AI initially tend to automate entire processes. While this is theoretically better because it frees up the human entirely, it often leads to workflow disruption, a lack of trust, and low adoption among healthcare professionals. This is especially true for high-risk tasks directly impacting patient care.

A healthcare professional, who is accountable for the care given to the patient, is acting properly when he decides not to adopt an automation tool that he can’t inspect nor understand.

A much better approach in these situations is to add AI as an assistant, augmenting healthcare providers’ existing workflows and enhancing their capabilities without removing them from the loop. This strategy not only builds trust but also ensures that AI tools support and empower users. By integrating AI thoughtfully into the workflow, healthcare professionals can leverage advanced technologies to improve patient outcomes while maintaining control over critical decisions. This requires a robust user-centered design process, ensuring that AI applications are not only functional but also intuitive and aligned with the practical needs of their users.

Strategies for Improvement

To bridge these gaps, a more integrated and collaborative approach is needed. Stakeholder and end-user involvement in the design, development, and deployment of AI solutions is crucial. Engaging healthcare professionals, patients, and administrators in the design process ensures that AI tools meet real-world needs and integrate seamlessly into existing workflows. Additionally, focusing on user-centered design principles can create more intuitive and effective AI applications.

This involvement will help identify internal champions who will use and advocate for the AI technology as early adopters. These champions help facilitate the integration process and ensure that the technology meets the practical needs of end-users within the healthcare system.

Once you’re ready for deployment, pilot projects are an effective way to introduce AI in healthcare while continuing to refine your application with direct feedback. Offering a trial period, such as six months with reduced fees, enables practices and health systems to evaluate the AI’s fit within their operations before full commitment.

Alvee: a case study of successful integration

Alvee is an AI-driven health equity data management platform that helps practices and health systems take every opportunity to advance health equity and anticipate their patients’ social needs. Instead of reinventing the wheel, they started with EHR plugins. They figured out their way to get into the care navigators’ journey, and they then used AI to enhance their work when assisting disadvantaged people. 

A series of augmentation UI patterns, where users interact and collaborate with AIs makes the best of the two worlds: the AI helps summarize notes and patient information and even proposes a care plan with linked resources, covering much of the tedious work. However, the care plan is then reviewed, adjusted, and approved by humans involved in the process. Providing control and agency over the final result makes users feel accountable for the proposed plan and offers more opportunities to embed feedback loops that can continually improve the underlying ML model.

You can read more about Alvee’s AI app design and how we helped create an AI application that enhances rather than automates experts’ daily tasks.


In conclusion, while AI offers tremendous potential for transforming healthcare, its success depends on addressing the disconnections between technological capabilities and user experiences. By focusing on integrated, user-centered design and involving stakeholders throughout the development process, we can create AI solutions that truly enhance healthcare delivery, improving treatment outcomes and operational efficiency.

We have a process that can help you detect opportunities for AI integration, understand the gaps within existing tools and the reality of your end-users, design solutions tailored to the actual process, and perform an initial validation. It’s all packaged in a fast-paced format that will significantly enhance your proposed solution without wasting time. Meet our AI Design Sprint, and reach out to us at for a direct consultation!