Empowering healthcare and infrastructure through innovative Generative AI research

Opinions expressed by Digital Journal contributors are their own.

Legitimate concerns about the development and implementation of Generative Artificial Intelligence (GAI) include privacy and security of personal information, a lack of explainability, transparency, and accountability, and ethical implications that can cross protected boundaries. While any technological advancement always comes with its own risks, it’s hard to deny the numerous benefits and transformative potential GAI brings to various fields.

Addressing these challenges would require a comprehensive approach involving policymakers, industry leaders, and researchers such as Dr. Shivani Shukla, a Generative AI expert who has created and implemented multimodal language and vision models while mitigating risks associated with the technology through ethical and value oriented practices in the absence of overarching regulations. She has led projects for companies in the healthcare and infrastructure industries with the outlook of avoiding the misuse of GAI through better practices incorporated in the design phase of product development. Dr. Shukla’s perspective on GAI emphasizes its positive impact and practical applications in society, rather than viewing it as a technology that will overpower the world. While the regulations are underway, she believes that self-regulation can go a long way.

Adapting a better perspective for infrastructure

GAI systems can produce precise and detailed 3D infrastructure models that can then be utilized for a variety of purposes, including examining a building’s structural integrity, planning renovations or expansions, and simulating the impact of natural disasters. They outperform 2D images that lack detail and realism, more so since they cannot always provide a holistic view of the physical world.

The presence of 3D models enables designers to analyze different design options and identify clashes or conflicts among architects, engineers, clients, and other stakeholders before any construction begins. This helps streamline the design process, reduce errors, and minimize costly changes during construction. Moreover, it provides a powerful toolset for complex design concepts that can be easily conveyed through visual representations. 

This leads to better understanding and communication among everyone involved.

Within Building Information Modelling (BIM), a process used in architecture, engineering, and construction, GAI enables the generation and management of visual representations of infrastructure. Along with 3D perception, such models can further enhance the efficiency and accuracy of infrastructure assessment. Dr. Shukla has led a project for an infrastructure data sharing and enterprise management platform to help build 3D reconstructions from two-dimensional, drone-captured images for infrastructure assessment. This entailed the use of her expertise in AI models, Statistics, and Optimization. Incorporating her knowledge of GAI with 3D perception into BIM has further impacted the company’s analytical and solutions-based decision-making processes

Developing innovations for global healthcare

GAI holds significant potential in revolutionizing healthcare. The industry is long overdue for a technological upgrade, especially in improving patient outcomes and enhancing diagnostics. AI has become a powerful tool in the healthcare industry, with providers using GAI models to accelerate the discovery and development of drugs. While it is still not a replacement for traditional drug discovery methods and human expertise, it has been used to reduce costs and efficiently generate new data instances that resemble training data. In terms of medical imaging, AI can read and evaluate medical images, which leads to a significant reduction in triage time of urgent cases and improves the allocation of these images to radiologists. On top of that, GAI can generate synthetic medical images that closely resemble that of an actual case. This can be valuable in situations where representative data are scarce. Synthetic images can also be used to train machine learning models, augment datasets, or simulate rare or challenging cases.

By analyzing healthcare datasets containing patient information, medical records, and test results, GAI algorithms can identify patterns, make predictions, and assist with prognoses. These personalized treatment recommendations can improve resource allocation and healthcare outcomes for all stakeholders involved. While current regulations in the medical industry prevent the use of patient data or any sensitive data to train GAI models, Dr. Shukla is leading the efforts of a cutting-edge healthcare app to include GAI into their technological platform. The app facilitates conversations among healthcare providers to enable better care in rural areas within the United States and has plans to expand internationally. She is developing large language models specifically designed for the healthcare context to summarize the data and feed it into subsequent AI models that can identify patterns to make sophisticated predictions for improving patient outcomes. The models that Dr. Shukla is building for the company, will lay the foundation for their future capabilities in terms of multi-modal data synthesis, personalized medicine, and real-time decision support.

Viewing the long-term positive effects

Dr. Shukla sees that the use of AI is still in its growth phase, and rather than racing to become leaders in GAI, healthcare and infrastructure businesses must view its long-term implications for their industries. As her intensive academic research and projects with significant companies continue, Dr. Shivani Shukla remains a proponent of the technology. She’ll continue to develop and implement it for objectives that will empower the healthcare and infrastructure industries and benefit people in the US and potentially more countries.