Hello everyone. I'm Sai Deepak. I'm a staff software engineer at Google. I lead the Maps quality team, where we deal with a wide range of challenges related to auction algorithms modeling advertiser conversion quality, and the overall quality of ads. I did my bachelor's at Indian Initiative of Technology and after graduation I started working at Google. I initially worked on the business software and then for a year and a half. And then from then onwards, I have started I've worked on the map sets. Today I'm excited to talk to you about leveraging artificial intelligence for medical cost, transparency, and consumer empowerment. I think this is a topic I'm deeply passionate about, as I believe technology can play a transformative role in making healthcare more affordable and accessible to everyone. And I'm pretty sure everyone in their life, Had to go to a doctor, had had a medical emergency, had to had to look up costs for either a procedure or or a medication. And this is actually a pretty big issue in United States and even in other other parts of the world, where people can get pretty big surprises on the cost. In this in this in this in this article we have explored like how we can use artificial intelligence and how we can try to resolve this resolve the issue of consumer awareness. Yeah, let's face it. The U. S. healthcare system is extremely complex and often opaque. It's it's whole working is not even understood like by the doctors and certainly not by the patients or medical care providers or anyone. The patients frequently find themselves in the dark when it comes to when it comes to understanding medical procedures and the cost of medical procedures. This lack of transparency often leads to a significant financial strain for millions of Americans, but also results in delayed or avoided care. And it, it the the delayed or avoided care can lead to making the issue even worse and driving up the long term costs. For example, one of my friends had when he was studying, like one of my friends had an injury in his leg, but because he was he, because he did not have a good medical insurance at this time, at the time he delayed the care till till he graduated and until he was able to get a job, but that delayed led to exacerbation of the injury to his leg. And ultimately ultimately long term long term issues and the long term costs. As a professional deeply involved in the intersection of the technology and healthcare, I'm hoping to use cutting edge innovations to solve these critical challenges. Today I'm excited to announce like a AI driven platform that has a potential to revolutionize the healthcare affordability and transparency by empowering customers with much more accurate and much more personalized cost information. This platform uses the the newest AI techniques such as like artificial intelligence, ability to process like large amount of textual data. And integrating them with analytics and it it changes how patient interacts with the healthcare system and it can improve the quality of the the care and make the information more accessible and affordable for all. There are various problems in the problems in the healthcare health healthcare transparency. One is the high and another is the unpredictability of the cost. Basically the pricing is very opaque. Sometimes the insurance covers it, sometimes it don't. Like sometimes one one provider charges one one price. Sometimes another provider charges another price. Sometimes the prices can change with zip code and the cities leading to unexpectedly very high medical bills, lack of transparency. Consumers don't have a clear way or dashboard or a software. To understand like the clear and accurate cost for the medical procedures. This leads to a significant number of financial stress and poor decision making in the absence of like complete accurate picture to the users. Financial strain the burden of the medical costs like often leads to delaying or completely avoiding the care together. And we have various studies which shows that for example, like having regular care or the or the yearly checkups like can actually reduce like the the much the long term costs. Much more significantly inefficiencies. The inefficiencies in the market contributes to like inflated prices and reduce competition among providers One of the biggest cause for this is just the lack of transparency in this area where where very few people have access to the prices unlike in other systems, for example If you want to buy clothes or if you want to buy buy any electronics, you can just simply go to amazon. com or google. com or any other, like any other aggregator websites. And you can simply look up like the properties of all the electronics goods. What is the price? What is how much does it take to deliver everything? Everything is everything is pretty pretty well exposed to the users. So they can take all this information and make the decision for themselves. However, in healthcare, like everything's very opaque. All this information is hidden away and known to very few people. Comparison difficulties. Obviously this lack of transparency makes it very challenging to compare the costs and the quality across different healthcare providers, limiting their ability to make an informed, educated choice. Solution. Our AI driven platform has three has has three pillars. One is the AI power transparency. So our platform uses newest LLMs and machine learning to deliver accurate personalized medical cost estimates. We have integrated data from multiple sources, including hospital charge masters, insurance claims, and user reported costs to create a comprehensive pricing database. This way we have used LLMs to basically like ingest the data and pass the data and Or put the data in a much more formatted way. And we use this we can then use this use this data and run analytics and give personalized information to the user. Real time estimates. The system provides like real time, like personalized cost estimates within seconds using all the data that we have extracted. And it is tailored to individual specific situation, individual's area, and the insurance the coverage they have. How it works. We have an A Web powered cost estimation process. So first we collected the data from various aggregators, either hospitals or the insurance claims, or the user reported. A user can also themselves report cost of the cost, and using all these various sources, pretty robust pricing database. And then we with respect to the machine learning algorithms, we have like advanced machine learning algorithms where it takes all this data and we train models on it, with various features such as regional variations, like provider specific pricing, user specific insurance coverage, and various various other other signals to train a model to to understand like the the, to understand the landscape of the pricing real time estimation and based on the model we have trained, we can provide like a cost estimate for for the user with based on based on their own issues, their own geographic area, their own insurance provider and validation the real time data validation techniques ensures that the estimates are correct and accurate. Okay. And the user can themselves give feedback to us and help us in iterating and improving the the data output we provide. Some of the key features of the platform are the personalized cost estimation. Giving the personalized cost estimation with with a variance of plus or minus 50 percent for the common procedures and plus or minus 30 percent for the complex procedures. This allows the user to plan for their healthcare expenses more effectively. Even though this this pan might seem pretty big, this is actually much, much better than the current blind the current the current Peace out. The current complete lack of data that exists. This at least like users this at least gives an estimate to the to the users. And the huge thing is provider comparison. It enables users to compare like prices and the, the quality ratings across multiple healthcare providers. This significantly increases the competition among them and obviously like people can take the prices and the quality in order to make that decision to whether to go to the healthcare providers. AI provides negotiation support. It uses assist users in negotiating lower bills creating creating letters, attaching information, attaching the billing data, attaching the comparison data, and the users can leverage to to identify the areas where the cost can be reduced. On average some users saw 18 percent, more than 18 percent reduction in their medical bills by using these the generated support information. And we have to we obviously want to make sure that there is a fairness and transparency in the cost estimation. We have a significant number of mechanisms in place to mitigate bias and ensures that all the reasons all the users receive equitable treatment and we don't use any signals which can distinguish between for example way where we distinguish between two users. This in this slide, we shared the impact and results from the impact and results from this. And we have seen significant cost reduction for the users who have used the negotiation support feature. They saw 18 percent reduction in the medical bills and significantly decrease low out of lower out of market expectancies. You also see an increase in improved access to care. Now that like users have like an estimation of like how much How much it will actually cause, they can do a better budgeting of the finances and and actually and actually use, and actually try to get the cost sorry try to get get the care and budget for it. And users are now more confident in their ability to afford the necessary treatment. 20 percent reduction in the avoided or delayed care. This this this also obviously led to like more better health outcomes. Reducing these financial barriers people's health outcomes have improved, they're able to save more costs in the longterm. They're able to have a more healthier, happier, and healthier life. And we hope that this will have a significant impact in the market. This has the potential to disrupt the healthcare by increasing the competition among the providers, driving down the overall healthcare costs improving the transparency in this area, reducing the anxiety that is faced by everyday people. There are significant challenges while we were doing the implementation of this. One, I don't think this would have been possible without the breakthrough technologies that have come in the last two and three years. For example, the context for example, analyzing the textual data. And generating more structured output from the textual data would have an extremely hard without the use of without how good the large language models are now, which can pass through hundreds of pages of the document and output the structured data that we are accurately asking for. With respect to the data and the privacy we use advanced encryption techniques and secure computation to product user data. Differential privacy measures ensure that individual data points cannot be traced to the specific users. In fact, we don't even keep any individual data. as soon as we train the data is just thrown away. With respect to the data accuracy, we use a federated learning. It allows the platform to continuously improve this algorithm without compromising users privacy by mainly not taking any user use of personalization signals. And the real time data validation ensures that the estimates are based on the more current and more accurate information available. So to do this, like our models train trains continuously on on the latest data on the latest data points. And we give we we force the model to give like lower and lower, lower weightages to the to the older data. So that like for example, the prices are based on like the 2023 and 20, for example. The cost estimates for 2024 are based on 2023 and 2024 data, not based on the two 2016 data. The model like automatically uses the old data for generalization, but it uses the more new data for giving more accurate data points to the users. I think ethical AI, we have a robust ethical AI framework in place to ensure the fairness and transparency in cost estimation. We don't take any we don't take any users identifying input features or try to differentiate between two people. The platform actively works to mitigate the bias and ensure equitable treatment for the all users. This, we mainly do this by not having any users any user personal IIS identifiers and treating like two people with the same with the same condition and in the same way. Okay. Here I want to go a bit into the future and enhancements. So we wanted we wanted we have a few other ideas for the future that we would like to do. We want to integrate with the electronic health records. This way we can provide like a more comprehensive view of the patient outcomes. And the patient costs, it allows for the personalized long term financial planning. This data can be integrated into the other the financial tools. And this can this can let people, for example, plan their retirement, plan the costs that will plan the general costs that they will have in the long term. And and also more help them in the saving for the long term for the health health costs. We want to have predictive health spending models. This this platform will develop predictive models to help users anticipate future health care costs based on the medical industry and lifestyle. This will lead to enabling better budgeting and the financial planning. For example, let's say someone has diabetes. This this model can use that information to know for example, what is the cost on average they will be using they will be seeing one year from now, five years from now, 10 years from now, and during the retirement. What kind of what is the rough estimate of the medical cost they they might see? What, how much is the, how much is how much money they should save right now? And it may, it can even help give suggestions to to reduce and reduce the amount of money that is needed. Feature expansion currently the ongoing development is focusing on expanding platform capabilities to cover more complex processes and broader datasets, ensuring it remains relevant and useful across a wide range of healthcare scenarios. With respect to the market opportunity, we see a significant the potential in the market and the growth. Obviously, this is like a, this is like a multi trillion dollar multi trillion dollar area where even if we can make like a 1 percent dent, we'll be very happy. And that'll be a significant impact on both in terms of the costs saved in terms of the anxiety saved by the by the users and overall improvement in the health and happiness of the users. There is significant growing demand as the healthcare cost continue to rise above inflation, there is significant demand for tools that can provide like cost, transparency and help and consumers to make informed decisions. This data will help everyone like this, will help consumers as will help businesses. This will help everyone to make a more informed decision markets. As the market for AI driven healthcare solutions is projected to grow significantly. Especially with the new technologies such as large language models, which which can which can understand like the medical information much better, which can which can transcribe the medical information and generate like more actionable information. So there are substantial opportunities for these platforms to focus on the cost transparency and consumer empowerment. Consumer competitive landscape we are acknowledging that our platform, it's it's not unique. It stands out in the crowded market due to its unique combination of the personalized cost estimate, provided compassion and AI powered negotiation support. However, we do want to do well and give better service to the users so that this can help them in their life. With respect to the partnership, we would like to have forged a strategic partnership with healthcare providers, insurers, technology companies, and also to create new streams and expanded reach. To conclude the cost of healthcare has has has risen significantly like about inflation. And this is one of the significant significant financial financial issue that people people face with. And this is one of the significant things that people fail to account for account for their in, in their life. And this leads to a significant number of anxiety significant number of significant amount of impact in their life in the, during financial and budgeting and so on. There is a significant need for the transparency and consumer empowerment because this this area is extremely opaque. Extremely unknown and extremely complex and bureaucratic, unlike many of the sectors, for example, like financial sector consumer and electronic sector services sector, where there's a significant number of you can just go and look up online to get all the information. This is extremely extremely opaque and this is screaming for more transparency and openness. The AI driven platform addresses these pressing issues by providing users with the accurate personalized medical cost estimate. And it helps them to make informed decisions about their health care. And our project also would not have been possible without the The large language models that have come in the last two and, two and a half to three years. Because most of the, most of our work involves transcribing like large amount of textual information. With which, which also includes the cost of specific procedures, the cost of pharmaceuticals, and so on. This would not have been possible to do with with the previous with the previous textual understanding. But with the new LLMs, such as with the new the new, with the new advanced LLMs this becomes much, much more easier. We also integrate the data from the diverse sources, including charge hospital, charge masters, insurance claims. And also we also let the users themselves report the cost and our platform creates like a comprehensive pricing database, which offers real time insights into all these medical expenses. This level of transparency is designed to elevate the financial stress that so many Americans face when navigating the complexity of the healthcare system. We hope that with this project we make we make impact and let let consumers understand the cost of the healthcare cost of the procedures, cost of the general cost that they are paying and make. more informed decisions in their life. Feel free to let me feel free to contact me if you have any further questions and thank you for your time.