Hi everyone. I'm Amala Han. I work as a principal data engineer at a management consulting firm. I have over a decade of experience in leading the data architecture and digital transformations, building scalable and robust data pipelines at an enterprise scale. Thank you for joining with me Today we are going to talk about how generative AI is reshaping the healthcare insurance workflows, particularly in claims processing and fraud detection. This is a space undergoing massive transformation, and gene AI opens a door to possibilities we couldn't reach with traditional automation. Let's dive into the topic to begin with, let's start our problem. The healthcare ecosystem is massive. Tens of millions of claims are submitted, reviewed, processed and cleared every day across several payers, providers, and insurance companies. Claims processing are still heavily manual. People reading clinical notes, checking codes, line by line, and the data. It's everywhere, like in papers and portals. EHR systems attachments that does not talk to each other. Fraud detection is also based on rigid rules that don't evolve fast enough. All these issues cost the healthcare industry over a hundred billion dollars annually. Combine all that with other regulations every year. We have hit a point where complexity is outpacing the human capacity. We need a technology that can understand the context, learn the patterns, and scale, and that's where Genea comes into picture. Genea is revolutionizing the healthcare industry by establishing scalable, secure data architectures and EA models that power the clinical and operational advances. They also enable transformative claim processing by automating the complex adjudication processes, reducing manual errors, and accelerating the payment cycles. Advanced gene AI models strengthen the fraud detection through patent analysis, anomaly identifications, and predictive alert on suspicious claims driving cost containment, and system integrity. Responsible AI implementation also ensures that gene AI deployments are transparent, fair, and compliant with ethical standards. Supporting trust and accountability in care delivery. In practical terms, gen I tools are already enhancing productivity and outcomes across care continent driving measurable improvements in efficiency, accuracy, and patient experience. This isn't just efficiency, it's a reinvention of how insurers operate. To make Janea Ashley useful inside an organization, it needs structure. This five layer architecture brings order to the kiosk, the data ingestion layer. This is where we ingest the data and pre-process them, and then we understand the clinical language, leveraging the natural language processing and validates it with medical standards, make intelligent decisions and automate the workflows end to end. This is how we get intelligence into the operational core. Let's look at each one closely. The first layer is the data ingestion and pre-processing. The biggest barrier in any data pipeline is a data fragmentation. Claims come in several formats, like PDFs, hhl seven messages, or even DCOM messages, images, generic consumes data in all of these formats and converts into structured data for downstream processing. The ingestion pipeline performs data cleaning, d deduplication, and enrichment as part of the data. Pre-processing steps. Think of it as laying a solid foundation. Without clean unified data, no AI models can deliver reliable outcomes. Once it, once we have the clean data, the next challenge is understanding what that data actually means. That's where the NLP takes over. Many of us would've heard or even seen medical data, and it's not simple text. It's full of medical codes, aggravations context, and clinical nuances. This second layer leverages the NLP to extract meaning from these codes and a observations gene I reads, physician notes, discharge summaries, and claim negatives. Then it identifies entities like diagnosis codes, procedures, and medications. It doesn't stop there. It goes on to build contextual understanding. For example, it can distinguish between a follow-up visit for diabetes versus a new diagnosis of diabetes. There is a subtle difference that affects the building accuracy, and the result is a claim that's enriched with semantic intelligence ready for automated de positioning. These remaining layers focuses on validating the correctness, which means the goal here is not to reject the bad claims, to improve the equality before it, before the submissions, and then detecting risks and automating the smart routing. The final orchestration layer ties the entire ecosystem together. Claims that are high, confidence can auto approve and questionable claims get escalated. Humans focus, humans can focus where human judgment is needed and everything becomes faster and more consistent when it comes to processing the medical claims. Transforming this process with gene AI in healthcare is a game changer for day-to-day operations. By using a advanced AI models to read, understand, and automate claims documentation, healthcare organizations can dramatically cut down on manual paperwork, streamline the approvals, and reduce errors that often slow things down. Gena also helps spot inconsistencies or missing information instantly flagging claims that need attention and freeing up the staff to focus on cases that really require human judgment. This means claims get processed much faster. Sometimes in minutes instead of days leading to quicker reimbursements for providers and less waiting for patients. On top of that, AI driven automation brings transparency and removes the guesswork from the whole process. Teams gain better visibility into claim status, bottlenecks and operational metrics, allowing for smarter resource allocation and continuous improvement. In short, Genea delivers significant efficiency gains that save costs, boost productivity, and enhance satisfaction across the board for the patients healthcare workers and payers, all alike. Now, let's talk about the real differentiator, the fraud detection, how fraudulent claims are captured in traditional systems. The answers to this question depends on codified static business rules that capture the thresholds like. Flag the claim if an amount is greater than X dollars or duplicate claims within 30 days. However, frauds evolve so quickly, often just likely changing patterns to bypass those frauds. January brings multimodal analysis. This means it combines the structured data, claim narratives, and behavioral signals simultaneously in order to detect the suspicious activity proactively by integrating these different data types. The system gains a more comprehensive understanding than any single source could provide. Leading to more accurate and efficient claims education, it can identify sophisticated fraud patterns even before human auditors. Notice them and send notifications to the right personnel to act upon. For example, the most complex and hard to detect patterns are often buried in structured data, unstructured data such as clinical nodes, doctor's letters, and claim description. NLP could actually reveal a hidden pattern in these unstructured data. Multimodal fraud detection in healthcare is getting a major upgrade thanks to the advanced AI techniques that pull insights from different kinds of data to spot suspicious activities more accurately. With graph neural networks, the system maps out the relationships among claims patients, providers and transactions. Think of it like building a social network for claims data. If certain providers, for example, are unusually connected to clusters of questionable claims, the network highlights those as red flags right away. In temporal pattern analysis is all about watching for the odd timing the AI looks at when procedures, prescriptions, or billing events happen and can quickly spot if someone's submitting claims and patterns. They just don't make sense, like frequent charges at strange hours or burst of identical procedure, which is a classic sign of fraud. Document Authenticity assessment goes further by letting AI review paperwork, invoices, and digital documents for signs of tampering or forgery. Gene AI can detect subtle differences in wordings, formatting, or signatures, alerting the investigators to documents that look doctored or generated by bots instead of genuine human submissions. Last but not the least, anomalies, scoring models, crunch all these clues and assign risks. To each claim on transaction when claims stand out statistically, whether due to the networks timing or document content, the system prioritizes them for review. Ensuring team tackles truly suspicious activity first and don't get bogged down by false alarms. Together with these techniques, meet faster, smarter, and more comprehensive fraud detection. Making it much harder for bad actors to slip through the cracks and helping healthcare organizations protect their resources and patient's trust. This gives insurers a 360 degree defensive shield against fraud. I love this one. The synthetic fraud scenario generation AI can create water fraud scenarios that haven't happened. This lets insurers train their systems. Before a new tactic shows up in production, we move from reactive to predictive defense. Here it's like having a heads up on future fraud. Using generative modeling, we can simulate fraudulent claims to stress test the detection pipeline to protect against future threats. This approach helps the system to learn from possible future fraud events even before it occurs, thereby improving their resilience, adaptability, and system integrity. Talking about responsible AI in healthcare, accuracy is important, but trust is essential and responsible AI is not optional. It is the foundation of adoption. We must keep all the personal information, the PHI information private and we need to explain the decisions clearly. We need to monitor for biases, and we also need to comply with the HIPAA and other audit requirements. Explainability is equally essential decision makers and patients should be able to understand how AI reached its conclusions, whether about claims denial, or risk ratings or fraud alerts. Responsible AI frameworks encourages transparency, feature important analysis, and ongoing testing so that any biases are supported and corrected by com. By combining the compliance and explainability insurers, build trust and accountability. Which strengthens the relationship with your customers and support better fairer outcomes in the healthcare sector. Organizations in healthcare face big hurdles when rolling out ai, especially when it comes to data fragmentation. Keeping models up to date and helping the workflow adapt. New technologies, it's not about just technology, it's people, process and change data silos, workflows, readiness, and model drift. All of these require strategy. Data fragmentation is a classic challenge. Health information is scattered across electronic health records, insurance, databases, labs, and even old paper files. This makes it tricky for AI systems to get a complete picture, potentially missing important signals or even causing redundancies. The solution, investing in secure data integration platforms and robust interoperability standards so that different systems can talk to each other, and AI has one clean, unified data source to work from. Model maintenance is another roadblock. AI models need regular updates to stay accurate as new diseases emerge. Regulations change or patient populations shift, many organizations struggle to set up automated monitoring and retraining pipelines. The best fix is to build strong ML ops team and use continuous monitoring tools, making sure models are retrained with fresh data, evaluated for biases and version for accountability. Then there's workforce transformation. Staff needs training and support to trust what AI is doing and uses. It uses its results effectively, but fear of job loss or lack of technical skills can cause resistance the way forward is ongoing Education. Change management and bringing clinical, operational and tech teams together so everyone's included in the AI journey. Upskilling programs and transparent communication make the workplace more resilient and ready for future. By tackling these challenges head on healthcare organizations can unlock more value from ai, improving the healthcare, and empower their teams for what's next when implementing the AI roadmap. Jena in healthcare insure space, breaking the journey down into four easy step phases. The first phase is a foundation. Think of this as laying the groundwork. Teams focus on setting up secure pipelines for collecting and handling data, picking and configuring the natural language processing models, and building initial validation frameworks to make sure everything runs according to existing business rules. The next phase developers get fraud detection models up and running. But start in shadow mode, managing the models alongside the current systems, not replacing them existing ones just yet. This phase also involves creating helpful explainability tools for claim adjusters and setting up AB testing infrastructure. So model performance can be accurately evaluated in the integration phase. Gene systems are connected to core insurance platforms. Workflow orchestration gets attention with automated routing and escalation for claims processes. Monitoring dashboards are also introduced, so stakeholders have visibility into the key metrics and can keep an eye on model performance in real time. In the optimization phase, the focus shifts to continuous improvement. This includes building pipelines for ongoing learning, expanding AI coverage to more claim types. Boosting capacity for handling more claims and crafting advanced features like synthetic fraud generations and federated learning to make the systems smarter and more robust for the future. This roadmaps makes complex AI implementation feel manageable by breaking it down into stage by stage priorities, ensuring progress is measured, and teams know exactly what to tackle next as they move forward. So here's the big picture. Healthcare insurance is entering a new AI era and is incredibly exciting. Here are the main takeaways and practical steps for leveraging the gene AI in healthcare insurance context. All in a clear and structured way. Architectural thinking genea isn't about algorithms. It starts with a solid foundation. Five layer architecture that we talked about ensures a structured rollout, balancing cutting edge innovation with all important regulatory requirements that keep data safe and systems in check and beyond automation. Gen A goes way beyond that. It brings new capabilities to the table, like improving fraud detection, risk assessment, and making smarter decisions, all things that move business forward and add major value and success with AI means doing things ethically. That means focusing on explainability, protecting privacy, minimizing bias, and staying compliant with all relevant laws and standards. To really see the benefits, the organization should also start small and focused. Pick use cases that they can show clear incremental value, then build up bigger technical capabilities as expertise and their confidence grow altogether. These points offer a roadmap for organizations work, wanting to make GI work for them in the responsible, effective, and forward thinking way. Thank you so much for your time and let's together explore and experience the advancement of technology for the good cause. Have a good one.