My name is Ram Krishna. I'm working in a medical as a solution architect. Today I'm going to present a reactive to pre predict to transform, transforming medical device reliability through Secure, secure I architecture. Okay, so the traditional approaches whenever healthcare oration typically responds to the medical devices failure after they occur in any medical failures that is leading to the downtime and the delayed diagnostics and increase the operational cost. So this rare two parum creates a inefficiency in a resource allocation and impacts the patient's care delivery. So our vision is okay, is a data driven predictive maintenance. So that means, so the reactive system so we'll fix that. Any medical device values, whenever it is phase, then we'll apply the fix. And also it's a schedule maintenance. Whereas our reactive processes, predictive processes is a datadriven intervention and also automated optimization. So that means we're transforming the react to responses to prior two. So through IT enabled strategies and anticipate failures before they impact on clinical operations. The Secure iot Architecture Foundation is completely like completed in three components. One is cloud platform, and so here we are using a very zero trust security principles. We are using MQ TT and HGTP protocols. So these two protocols you'll use the transfer the data from medical devices to the cloud computing. So then once, so we have, we transfer the data, then we do the edge computing. That means we'll apply the local pre-processing to reduce the transmission volumes while detecting a critical anomalies. And third one is HIPAA compliance. So that means, so this is a mutual transport layer, security authentication with certificate rotation and real time revocations. So that pipeline of architecture is in in five steps. One is a device telemetry. So that means we are continuously monitoring operational parameters from the diagnostics devices. And local analysis and filtering to identify the critical patterns and anomalies in that, the edge processing. And we securely transport the data from transfer to the, from medical devices, the cloud computing infrastructure, using a security protocol that is three TPS and MQD protocols. And we'll store that in the time series. So then we'll optimize the database for the telemetry data and advanced analysis for further processing. And finally analytics engine. So there we dub the my machine learning models to generate predictive insights and maintenance recommendations. So the zero trust implementation is a mutual TLS authentication for all device communications. So certificate rotation schedules with the automated management. So that means we will automatically roll out the certifications on schedule basis for the each medical device. And real time revocations, that means for the, any compromise credentials. So that means if you feel like if you suspect anything which is not working, then we will revoke that security certifications and under in a real time revocations capabilities. And last one is we also do the audit trialing for the compliance reporting. So whenever you make any changes on the medical devices, and we do have a documentation for the further the documentation for the auditing added trail for the further the processing and reporting. So the machine learning development is the feature engineering. So that means we'll transform the raw telemetry data into meaningful predictive indicators using a signal processing, static statistical analysis, and domain specific knowledge integration. And also we have a class imbalance handling. So address the unique challenge of rare failures, events in your medical device, data sets through specialized sampling and weighing techniques. And explainable predictions. So we develop the in interpretable models that meets regulatory requirements for software as a medical device into a controlled healthcare environment, continuous learning. So that incorporated the technicians feedbacks and adapt to revolving the device behavior patterns for further improvement prediction accuracy over the time. So the feature engineering processes is a signal processing, so we will extract the meaningful patterns from raw sensor data through a frequency analysis, filtering and transformation techniques and static analysis drive key metrics including trends, variance and correlation patterns that indicate device health status. And domain knowledge that is integrate a clinical and engineering expertise to identify the features which proven diagnostic significances. So the main key improvements are, or optimize the resource location through predictive analysis. So based on the analysis, and we can optimize the resource allocation for that. Fixes and automatic ticketing system integrates a maintenance and workflow streamlines the maintenance of the workflows. So whenever device can use occurs, our workflows will trigger to create the automating of system automated out the ticketing system. And also we develop the realtime dashboards. So visualizations enable that data driven decision making. So that means we can predict, add the based on the data visualizations and so we develop the reports and dashboards the stake, enhance the stakeholder communication across the clinical and the technical teams. So that is the major key improvements with benefits of this solution. Okay. So regulatory compliance framework. So we also have a FDA requirements. So address software as a medical device, regulations with the comprehensive validation and verification protocols details. Maintain a complete documentation of all the system activities, predictions and maintenance actions for the compliance reporting. And we also have a quality management, so we'll integrate with the existing quality management system framework to ensure continuous co and accessibility. So we implement the workflow in the five steps. One is the assessment. In this section, what we'll do is we evaluate the existing infrastructures and identify the critical devices for the iot integration. And second bit is architecture data design. So we love the secure iot architecture with the HIPA compliances and the Zero trust principles. Okay, so then pilot development deploy the edge computing units and establish a secure data pipelines for the initial device plate. So that means we will identify, we will roll out this initially only for the thousand devices and not for the, we don't roll out for the major clinicals and hospitals. And then once we roll out the solution, then we'll our model training, we'll collect the tele data and develop a predictive models. With the validation protocols and scale and optimize, expand to full device fleet and implement continuous learning mechanisms. So the future innovations and reaches directions is embedded diagnostics, so meme sensors which resides in the medical devices so that sensors integrate directly into the device for enhanced monitoring capabilities and rail them health assessment. Okay. Based on that MIM sensor data, we will do the assessments for each medical device and federated learning. So privacy preserving analytics across distributed device plates without a centralizing sensitive data. And these still twin technologies. So we develop the simulation based on the maintenance planning, using a virtual replicas for the scenario testing and optimization. So we have some simulators, exactly a medical device, so where we can use for the, for testing. And we can see, we can capture the, all the test scenarios and also you can use for the training purposes. And the success factor says the stakeholder engagement across the clinical, technical, and administrative teams. Okay. Comprehensive to training programs for maintenance personal and clinicians, the clear communication of benefits and measurable outcomes. Phase rollout implementation approach to minimum disruption. Okay, so continuous feedback loops to refine predictive models and workflows. Okay? These are the major success factors with our solution. Key takeaways for implementation is security first, so we'll established a robust zero trust architecture with the P compliances before collecting any patient adjustment adjacent data from medical devices. Start smaller scale smart. So begin with the A pilot deployment. As I mentioned, we roll out for the small and local hospitals like, hundreds and thousands on critical devices to validate the models and demonstrate value before enterprise rollout. So then we can test for so before rolling out, so we'll demonstrate on the local clinicals. MS continues learning that third one. So every we year, we built the feedback mechanisms that incorporate technicians expertise and adopt models to evolving device behavior patterns. Okay. And last one is focus on operational integrations. Ensure predictive insights translate into a actionable workflows through automated ticketing and decision support systems. So as I mentioned, whenever device value occurs, our workflows will trigger, require automated ticketing system and will use the two solution. Based on the, model what we developed. So our models will trigger and identify the air predict to a solution to the technicians for the, device fix. So that is the end of my presentations. If you have any questions feel free to ask here, and thank you so much for giving me the opportunity.