Jake Luo, PhD

My lab specializes in levering machine learning (artificial intelligence) and natural language processing technolgies to process, analyze, and model large amount of data (big data) to discovery new knowledge, find unmet needs,  and generate novel applications. 

 I’m the founding Director of the Center for Health System Solutions. We have an very active collaboration network though the research center and my lab. I also serve as the Director of the Health Care Informatics graduate program at the HIA department. 

See Luo Lab’s research and development areas below:

Voice AI-enabled mHealth Innovations for Chronic Disease Self-Management

Collaborating with Dr Hyunkyoung Oh, Dr. Min Sook Park & Dr. Sheikh Iqbal Ahamed since 2016, our project’s goal was to leverage voice-activated AI technology to develop an innovative mHealth solution, VoiS, aimed at empowering patients with diabetes and hypertension to manage their conditions more effectively. This NIH-funded initiative sought to address the challenges of simultaneous self-monitoring for these prevalent chronic conditions, aiming to improve patient outcomes and facilitate better patient-provider communication.

IMPACT

The VoiS app, developed and evaluated with input from healthcare providers, demonstrated significant potential to enhance self-management practices for patients with diabetes and hypertension. Feedback from healthcare providers, gathered through a structured assessment process, underscored the app’s usability, effectiveness, and the positive reception of its voice-activated features. Providers noted an 88% improvement in the app’s functionality and usability in its final version, indicating a strong endorsement of its design and utility in clinical settings. This feedback is instrumental in advancing our research towards integrating VoiS into clinical practice, potentially transforming patient engagement and self-management strategies.

INNOVATION

VoiS represents a novel integration of voice-activated technology with chronic disease management. Its development process, guided by feedback from healthcare providers, exemplifies a patient-centered approach to mHealth innovation. By offering a user-friendly platform for monitoring key health indicators and facilitating communication with healthcare providers, VoiS stands out as a significant advancement in the field of digital health.

CONCLUSIONS

The development and evaluation of the VoiS app illustrate the potential of voice-activated AI technologies to address the complex needs of patients with multiple chronic conditions. Our findings indicate that such innovations can significantly improve self-management capabilities and foster more effective patient-provider communication. As we plan to integrate VoiS into Electronic Health Records (EHR) systems, we anticipate broader implications for the future of personalized healthcare and mHealth applications.

Advancements in AI for Cardiovascular Health: Hybrid Explainable Machine Learning (XAI) Models for ECG-Based Risk Assessment

Collaborating with Dr. Rodney Sparapani and Patrick Noffke, our project embarked on pioneering the integration of artificial intelligence (AI) into cardiovascular health through the development of novel hybrid machine learning models, with a strong emphasis on explainable AI (XAI). These models leverage electrocardiograms (ECGs) for enhancing the prognostics of cardiovascular risks, focusing specifically on heart failure (HF) and cardiac arrhythmia (CA). The endeavor aimed at harnessing the capabilities of deep learning (DL) and ensemble learning (EL) to predict cardiovascular events from ECG data, significantly advancing the paradigm of personalized medicine through transparency and interpretability.

The core of our research was the creation of a hybrid AI model that combines the strengths of DL and EL, integrated with the principles of XAI. This approach ensured that our predictive models were not only accurate but also interpretable, allowing healthcare professionals to understand the reasoning behind the AI’s predictions. By making AI’s decision-making process transparent, we aimed to build trust and facilitate its adoption in clinical settings.

We constructed and tested several DL and EL models, incorporating XAI techniques to ensure that each model’s output could be easily interpreted by clinicians. This process involved rigorous analysis of ECG recordings and relevant clinical data from a curated dataset of patients. The models underwent extensive evaluation through cross-validation techniques to ensure their predictive performance, reliability, and explainability.

IMPACT

The culmination of our research efforts has yielded a groundbreaking tool for cardiovascular risk assessment that prioritizes explainability alongside accuracy. The hybrid AI model, underscored by XAI, stands as a testament to the potential of integrating diverse machine learning methodologies to improve health outcomes. It offers a more nuanced, accurate, and interpretable prediction of cardiovascular events, thereby empowering clinicians with advanced tools for early intervention and personalized patient care.

INNOVATION

This research represented a significant innovation in the field of cardiovascular health, introducing a novel application of AI for risk assessment with a strong focus on explainability. The hybrid model’s ability to analyze both the raw ECG data and a wide array of clinical information, and to present its reasoning in an understandable manner, marked a substantial advancement over traditional methods. This innovation not only enhances the predictive capabilities but also opens new avenues for personalized patient care, guided by the principles of XA

CONCLUSIONS

The successful completion of this project marks a significant milestone in the application of AI to cardiovascular health, with a pioneering focus on explainability. Our hybrid machine learning model offers unprecedented capabilities for risk assessment, paving the way for enhanced patient care and the advancement of personalized medicine. This research not only contributes to the scientific community but also holds promise for tangible clinical improvements, setting the stage for future innovations in healthcare technology, underpinned by the principles of XAI.

Empowering Patient Understanding through User-Centered Medical Report Design: Integrating AI and UX Principles

Feedback from patients has been overwhelmingly positive, indicating that these components greatly enhance the readability and usefulness of medical reports.Our research aim to enhance patient engagement and comprehension across the medical reporting spectrum, focusing on making medical reports accessible and understandable to patients. This journey has led us to design innovative report formats for both radiology and obstetric ultrasound scans, specifically targeting the second trimester. By integrating artificial intelligence (AI) applications and user experience (UX) design principles, we address the unique information needs and comprehension challenges faced by patients, empowering them to participate more actively in their healthcare decisions.

Unified Design Approach for Medical Reporting

Our efforts have culminated in the creation of patient-friendly reports, such as the Smart Patient-Oriented Obstetric Ultrasound Report (SPOUR) and a redesigned radiology report. These reports are grounded in extensive patient feedback, gathered from social media platforms and online forums, highlighting the demand for clarity, engagement, and reassurance. The redesigned reports stand out by incorporating AI-driven enhancements and UX design principles, offering simplified medical terminology, infobuttons for in-depth information, illustrative images and icons, and an organized layout for easy navigation. 

Methodological Insights and Validation

The validation of our designs involved interventional studies comparing traditional reports with our innovative formats. Participants, recruited through platforms like Amazon Mechanical Turk, were evaluated on their understanding, perceived ease of use, usefulness, intention to use, and aesthetic appeal of the reports. The new designs significantly outperformed traditional reports, demonstrating that our approach to integrating patient-centric design and AI applications can greatly enhance patient comprehension and satisfaction.

Conclusions and Implications for Future Healthcare

The introduction of these redesigned medical reports marks a significant advancement in patient-centered healthcare communication. By specifically catering to the needs of pregnant women and patients undergoing radiological examinations, we’ve shown that thoughtful design coupled with AI can dramatically improve patient understanding and engagement. Our findings advocate for the expansion of such approaches to all areas of medical reporting, leveraging technology to further empower patients and improve their healthcare experiences.