|
The use of Generative AI (Gen AI) and Agentic AI demands robust data governance and security, particularly because of the sensitive nature of PHI (Protected Health Information) and PII (Personally Identifiable Information), said Kodeeswaran Natarajan, SVP, Healthcare Delivery, Indium.
The basics of HIPAA law, HiTrust critical control points, and FedRamp assessment must be considered when platforms are being built. A governance framework built on validation, monitoring, regulatory compliance, and ethical standards is paramount, he added.
Multidisciplinary AI Review Boards strengthens data science, clinical, regulatory, and ethics expertise to evaluate the models. We see the future of healthcare with collaboration between humans and AI. With advanced AI models integrated across clinical workflows, the traditional diagnosis cycle will become faster, precise, and patient-centric, he noted.
Cumbersome administrative tasks and data handling which are the long-standing bottlenecks in healthcare delivery can now be largely automated through intelligent Human-in-the-Loop (HITL) systems. These systems strike a balance between automation and oversight, allowing clinicians to redirect their time and energy towards patient care, said Natarajan.
Earlier automation waves were led by RPA (Robotic Process Automation) and HPA (Hyper-automation). Currently, this is powered by Generative AI and Large Language Models (LLMs) which is seen to unlock a new era. AI is no longer just a back-office tool, it is stepping into the clinical frontlines. From augmenting diagnostic workflows and treatment planning to support pathology studies and early interventions through predictive risk models, AI is enhancing the full spectrum of care, he pointed out.
At the point-of-care, AI will act as a trusted partner increasing human judgment, improving clinical speed and precision, and reducing cognitive load. As these tools evolve, clinicians are poised to become AI-empowered healers. It is a healthcare system where technology amplifies compassion, and machines support rather than replace the healing touch of physicians, said Natarajan.
The healthcare industry needs autonomous AI agents instead of relying solely on passive copilots. This is because co-pilots require constant human input, autonomous agents can act independently. CoPilots can bring 20 – 30% efficiency in terms of time-to-market and cost. On the other hand, most of the automated processes in the healthcare industry bring around 70% efficiency. An Agentic AI helps make clinical and administrative decisions improving self-reliance to 95%. A few FDA-approved AIs are widely used to diagnose diabetic retinopathy, cardiovascular diseases, and orthopaedics, he said.
Generative AI serves as a foundational tool to extract insights from unstructured clinical and regulatory data. A physician’s system maximises imaging data in PACS, CCD, FHIR, HL7 format, and clinical notes. At least 70% of clinical and regulatory healthcare data is unstructured - free-text notes, scanned documents, research papers, clinical trial records, regulatory filings, documents and compliance forms. In the case of wearables, implants, or monitors could promote the in-home dialysis, and Agentic AI could prompt the command centre in case of any adversaries proving the indispensability of this technology, said Natarajan.
|