As the intersection of healthcare and technology evolves, artificial intelligence (AI) has emerged as a transformative force. However, to truly improve patient care, AI solutions must go beyond targeting efficiency gains and decision-support; they must also embody empathy and enhance the human connection in medicine. This is the philosophy that underpins the concept of human-centered AI, which prioritizes patient needs and respects individual experiences. Dr. Maaz Shaikh, Vice President of Product Management at M42, discusses the application of AI in influencing the future of healthcare.
AI’s role in modern healthcare
The integration of AI into healthcare is creating new possibilities for improving patient outcomes. From enhancing diagnostic accuracy to personalizing treatment plans, AI tools are reshaping how care is delivered. Yet, the ultimate value of these technologies lies in their ability to complement, not replace, the human touch.
Administrative support AI tools aimed at reducing the burden on clinicians are meaningful only if they truly remove tasks from workflows rather than requiring providers to learn and adapt to new systems. Predictive analytics tools, for example, offer immense potential by enabling clinicians to process large volumes of complex data and identify health issues early, fostering proactive and preventive interventions that alleviate both physical and emotional burdens. However, these tools must seamlessly integrate into existing workflows without adding complexity, such as extra screens, steps, or clicks, which can become barriers to adoption. By prioritizing usability and ensuring these solutions fit naturally into clinicians’ routines, AI can empower healthcare professionals to focus on patient care rather than administrative challenges.
Human-centered AI in healthcare refers to the design, development, and implementation of artificial intelligence systems that prioritize collaboration between AI technologies and human healthcare professionals, ensuring that AI supports and augments, rather than replaces, clinicians’ expertise and judgment. This approach recognizes the value of both AI’s analytical capabilities and human professionals’ expert contextual understanding, empathy and critical thinking skills in medical practice.
Developed by M42 in partnership with Core42 and Cerebras, Med42, an open-access Clinical Large Language Model (LLM), has demonstrated exceptional potential. Med42’s latest version scored 94.5 percent in a zero-shot evaluation of USMLE Sample Exam questions, which reflects its high performance in understanding medical information and data and its ability to provide clinically backed responses to medical questions. The availability of a cutting-edge and clinically reliable AI technology like Med42 can power multiple-purpose specific solutions for some of the most pressing global health challenges. These include reducing the administrative burden on clinicians, enabling healthcare professionals to focus on patient interactions and dedicate more time to building meaningful relationships with their patients and enabling clinicians to support patients in self-care outside of the clinical interaction with personalized evidence-based material.
Physician-centered design: A necessity for increased collaboration
Clinicians like physicians and nurses are the gateway for any healthcare innovation to reach patients. Clear benefits and seamless integration into existing healthcare workflows, respecting clinicians’ current practices while enhancing their productivity and decision-making processes are essential for the adoption of AI in their routine clinical practice. Equally important is the understanding of cultural and regional contexts in AI systems. In diverse regions like the Middle East, tools like Med42 adapt to cultural sensitivities, ensuring that treatment and self-care plans align with patients’ beliefs and unique circumstances. This commitment to inclusivity bridges gaps in care and prevents disparities.
Combining human and artificial intelligence (HI + AI) promotes synergy, allowing clinicians to leverage AI’s strengths in data analysis and pattern recognition while maintaining their autonomy and contextual understanding. This hybrid intelligence acknowledges the limitations of AI in complex situations requiring nuanced judgment and the need for human oversight. For example, tools like M42’s AIRIS-TB are improving radiologist efficiency by prioritizing abnormal scans for expert review. This system enables the screening of over 2,000 chest X-rays daily, a tenfold increase on conventional methodologies and thus enhancing the accuracy of image analysis. In addition, M42 is actively advancing products and solutions for clinical documentation automation, AI Physician Assistant, pharmacogenomics and multimodal analytics for personalized care, laying the groundwork for improved physician productivity, better treatment planning and outcomes.
Empathy: The cornerstone of human-centered AI
Human-centered AI demands empathy at its core. Designing such systems involves collaboration among clinicians, patients and ethicists to ensure fairness, transparency and respect for dignity. These principles have guided the development of Med42, which enhances patient interactions by equipping clinicians with actionable insights while reinforcing the human element of care.
Dr. Shaikh, who is an advocate for empathetic AI in healthcare, reflects this sentiment: “Empathy is the foundation of effective healthcare delivery. By leveraging AI tools like Med42, we can amplify the quality and intensity of the human touch in healthcare and ensure technology enhances, rather than detracts time and attention from, the patient-provider relationship. Improved interactions with patients will enable clinicians to make more holistic decisions, provide better patient experience and develop better health literacy and engagement in self-care outside of the clinic.”
Navigating ethical and privacy challenges
AI’s capabilities come with significant responsibilities, particularly concerning data privacy and ethics. Strict adherence to global standards such as the General Data Protection Regulation (GDPR) and the Health Insurance Portability and Accountability Act (HIPAA) is non-negotiable. Safeguards like encryption and controlled access are essential to maintaining patient trust and protecting sensitive information.
Transparency is equally crucial. Providers must have access and understanding of the recommendation process and references to evidence supporting AI recommendations. Patients must be provided with a full understanding of how their data is being utilized and have the autonomy to opt out at will. By fostering a transparent and ethical ecosystem, AI can build stronger, trust-based collaborations with healthcare providers and their patients.
Dr. Shaikh referred to the MEDIC Framework published by M42, which evaluates large language models (LLMs) across critical domains—Medical Accuracy, Ethics and Bias, Data Understanding, In-Context Learning, and Clinical Safety. This framework places patient safety at the core of AI development and implementation.
Making innovation accessible
M42 pioneers the MEDIC Framework, which evaluates large language models (LLMs) across critical domains—Medical Accuracy, Ethics and Bias, Data Understanding, In-Context Learning, and Clinical Safety. This framework places patient safety at the core of AI development and implementation.
The transformative power of AI must be accessible to all, irrespective of geographic or economic barriers. Initiatives like providing free access to Med42 for non-commercial purposes exemplify how AI can be democratized. M42 is also expanding its technological innovation efforts globally to ensure that even underserved communities can benefit from sustained AI in healthcare even with limited access to computing infrastructure and capital.
Looking ahead, advancements in safe ethical, responsible and explainable AI will strengthen trust in AI systems, enabling the use of AI for data-driven clinical decision-making as standard practice. Partnerships among technology providers, clinicians, and academic institutions can facilitate high-quality testing and real-world validation. Integrating multimodal data sources and developing AI models that understand this data —including genomics, clinical histories, and wearables—will further empower our providers and accelerate the realization of the vision for precision medicine for all.
Shaping the future of healthcare with AI
Looking ahead, the potential of AI in healthcare is boundless. From administrative support to predictive diagnostics to personalized treatments, AI will redefine every aspect of patient care. However, its success hinges on seamless integration with existing workflow and its ability to collaborate and support humans – fostering collaboration, empathy, inclusivity, and improving the quality of the human touch in every interaction.
