when the going gets tough the tough get going many of us have seen it felt it experienced it one such story i'm going to share with you my name is nivriti rai and i'm the country president for intel here in india as well as i have a functional responsibility i'm the vice president for data center products group i'm going to be talking about the use of ai for population scale solutions when the pandemic hit us early this year it was unprecedented it was prodigious and it was unfortunate what i'd like to tell you is that the world came crippling down on its knees this year what were the reasons why i'm calling it prodigious why i'm calling it unprecedented the reasons are three and all three get far more complex just because of the sheer scale the first reason who was impacted the citizens were impacted the enterprise impacted as well as the government was impacted the government's governance as well as administration as well as economics everything got impacted it spared no one the second reason why it's unprecedented is the why and the what the situation started rapidly changing and evolving no one knew what was happening there was lack of knowledge the complexity was going on increasing it started with wait this is a lung disease people with asthma you've got to be you know very very careful wait no it impacts the heart there's inflammation in the heart further there could be a liver impact there could be a kidney impact there could be a gastritis kind of impact it became so complex that the treatment option the doctors had you know was so difficult what to address what not to address who should be given this medication versus who should be given that meditation medication it became really really complex really really a wide set of treatments that were required so scale again was a problem here thirdly the third problem is what kind of treatments what kind of data where to access the data from there was lockdown there's data protection there was data movement regulation all those issues were coming up all those issues crippled the world to an extent people were scared there was fear there was data movement guidelines so the how became so so much complex and as the pandemic started you know to touch india two things came to my mind density of the population 1.3 billion people 20 percent of the world sitting in india and very dense second the gdp per capita of india is one fifth of the world while the world sees 10 000 per capita india sees two thousand so the challenge was if it touches india and it grows widespread like you know it was growing in other nations like italy it would be devastating for the country the challenge was this issue could be controlled and addressed only with the solution that is built on the backbone of technology technology was the only way we could come up with solution and that too the best of it intel india aws india and fracta and a host of other local partners started working together closely to build an open architecture-based solution open api locally hosted cloud-based all of this to enable multiple citizen-centric apps and augment the kind of work that both state government and central government were doing to manage this pandemic at the population scale i often use the word population scale because it's very befitting for the country like india the data is humongous the challenges are therefore you know multi-fold just to give you an idea the number of companies working with us were not just the three of us intel aws and fracta there were 14 to 15 other companies the alpha personalities you know we were working with everybody coming up with that a game everybody coming up with the best portfolio products that's another story another challenge which we could discuss some other day but bringing together 17 18 different companies all leaders all having great ideas was another challenge the scale the of the problem that you know we were dealing with was daunting so what was it requiring at that moment it needed the strongest most committed leadership it needed the smartest technologist and it needed the best best portfolio of products to address this pandemic at population scale what comes into your mind two companies aws and intel and that's where the two of us stood stall to drive the leadership and to come up with the solution that was required to address this pandemic we had already identified the requirement of this pandemic we had understood that it was going to be a cloud-based solution and it was requiring a scale that nobody had heard of intel often works very very closely with the communities and has a goal to become a responsible corporation our ceo bob swann has built a purpose for us and the purpose goes to say that intel will build world leading technologies that enriches and affects the life of every single person on this earth few words that stick out every single person and earth when i say every single person the one country you cannot miss is india it houses 20 of the world's population and when you say earth then you think enabling technologies and you think enabling technologies that help the environment one program that intel is driving is called rise it stands for responsible inclusive and sustainable technologies that can come together to solve global challenges global challenges that require cutting ed technologies leading technologies and are also enabling scale so looking at population we know that solutions like this uh require partnership i do not believe such large scale problems can be solved independently they have to be done in a collaborative manner collaboration with industry partners collaboration with academia as well as collaboration with government as intel india we have partnered on many such national level problems and come up with ways where we have an ability to come up with contextualized annotated data sets to build uh solutions which have unique algorithms as well as has support in terms of policy regulatory requirement the three focus areas that we have looked at is health we've looked at smart mobility and we've also looked at education over the last two two and a half years we've been collecting data from the road trying to come up with technologies for smart mobility india sees 17 deaths in an hour very unfortunate statistic so we've been collecting data and trying to analyze and come up with safety algorithms that can you know be saving these 17 deaths that happen on the road in terms of health we have come up with few diagnostics we've collected a lot of data on on boned uh degeneration we've collected data on lung cancer detection and we've all also used our technology to make the diagnostics faster for example a molecular dynamic simulation code called namd which we have used uh enables two times faster diagnosis we've used our avx 5012 technology to enable this i also must tell you that while you think of intel as a silicon company while you think of intel as a hardware company we have about 15 000 some software engineers who are working on all different kinds of optimizations if you think of of software we build embedded software we work you know to uh make sure that the hardware is optimized with the os all the way to application level optimization i must also tell you that we have solutions uh where we leverage for example uh you know library software libraries like openvino and we have enabled diagnostics which is 40 times 200 times faster by the use of these open source software for diagnosis of bone degeneration as well as cancer what does this tell us this tells us that over the last so many years technology has become critical to be able to solve problems at population scales and health being one what better time to push technology solutions uh than we are in the state now so the whole purpose of sharing uh this information is that we built the capability we have the portfolio of products and we have a direction coming from our ceo and coming from our leadership team to use cutting-edge technologies that enriches the life of every single person on this earth the intent is that this needs to become the dna of how we work why value creation economic uh gains is definitely our goal but while we do that becoming a corporate citizen is also extremely extremely critical for intel we what does this tell us now that we have portfolio products we have intent we have capabilities we have a hunger and we have a very complex problem to solve my next slide tells you that this problem which we have identified to look at how do we come up with a solution that can somehow you know monitor and create reports based on the understanding and the analysis of data that we get from this pandemic when we started our conversation with the government to flesh out what are the key tenets of this pandemic platform few things that came out that you know this solution this platform needs to be real time data is coming you know every day 24x7 we've got to make sure that the the dashboard that we display the reports that we create are real time this solution that we built has to have open architecture the variety of devices providing us data there's variety of uh different systems i know creating data various districts you know cities villages providing us data it had to be open in the way it was collecting data and the way it was architected 20 of the world population means that this solution has got to be scalable and flexible flexible as we get more and more learnings we've got to be able to change on the fly depending on the need of this hour i wanted to tell you just as a side note within intel we ourselves you know it's like eat your own dog food we realized that if we look at the big data analytics solution and flow we should implement it within intel this piece of data sitting here the customer there's pieces of data sitting in the factory there are pieces of data that you know are with the design uh guys whole lot of information sitting in different databases what we did is we leverage our own solution data analytics solution and i'm happy to tell you that in just last year just by the share of building a big data analytics solution within intel we were able to realize a 1.25 billion dollar business value that's how important understanding the architecture understanding the solution and then implementing is so most of you would have heard you know plan do check act but when we talk about ai based solutions we are talking about you know a workflow which starts with ingest you know ingest a whole lot of data you know from multiple sources prepare that data doesn't have you know standard formats you have to create and prepare your data such that your algorithms could leverage them effectively then obviously after you have prepared the data comes the analyze phase and eventually act that solution that you have built needs to be acted upon now all these processes and flows that i talked about are pretty damn challenging when you think of you know the very first ingest you're getting data from multiple sources you know multiple devices different regions perhaps even different languages the challenges are so vast that if you don't if you you blink your eyelid and miss out on certain change certain requirement you know the whole solution can fall flat a whole lot of contextualization annotation needs to be done before we start the analysis process this whole solution is not just simply model building experimenting but watching out addressing at every step you know the changes the flexibility the scale that is the requirement and finally it comes down to you know you have a a model that you have built you have uh an experiment that you have verified with now you can look at driving the action part remember that with ai the four things that become really really critical is we've got to make sure that you know our solution is not susceptible to bugs our solution does not have a lot of bias our solution cannot be manipulated or you know easily be infected by malware so bias bugs malware manipulation is something that we also had to look at and drive security as the inherent architecture feature once we drove all of this action part we were able to generate dashboards for government governance and administration for health institutes diagnostics and management as well as control and even looking at patients how do they manage and control some of the dashboards that we were able to create was risk hotspot mobility passport disease spread transmissibility um severity factors care readiness and many such things now what we were able to do is all of this was generated or planned or experimented on top of the regulatory and legal requirement that the country had so what we had is now a framework an experiment a poc ready what we needed is to work with the stakeholders and customers and look at how do we apply this solution and enable the customers to use it now we already have a wonderful vision an idea as to how to build the solution what we were able to accomplish is the the finalized solution vision and what were the technological requirements in terms of architecture as well as building blocks we also looked at customers and you know the key stakeholders and onboarded them we got them to contribute we got them to participate as well as heard from them what were their requirements we also leveraged the very rich prolific uh ecosystem of aws india as well as intel india and got them to participate with us in building this pandemic solution now intel and aws has a long history of working together we have worked on many a fantastic solutions which will be addressing the needs for today as well as scale to address the addre needs of the future andy jesse who's the cw of aws says intel is a very deep partner of aws and will be for a very long time that's not changing to add to it rahul sharma one of the leaders of aws india says that since we are connected at the hip navrati let's work with trust let's work with commitment and less drive innovation that is required to solve such population scale challenges in the terms of pandemic that we have now intel and aws have worked on many a partnership fundamentally intel believes that the hardware infrastructure that we leverage to build solutions on have to be you know open such that different partners different contributors could look at driving and building more innovations we could leverage their creativity many of the engagements that we have uh you know partnered with aws range from iot to uh you know i uh to uh developer uh leveraging alexa voice services to amazon echo show and so on one specific one that i want to talk about is the amazon ec2 amazon ec2 is the uh the compute the elastic compute that we have that uh you know in cloud that verbs web services could be based off this provides secure resizable compute such that you know security as well as the needs which are driven from dynamic changes could be leveraging this resizable capability that this solution has this met the need that exactly this platform required this pandemic platform required so we now had a platform on which the solution could be built we roped in many a partners to look at this cloud platform that we have and leverage it to build a solution that was open architecture i must tell you that the two states which were the customers for this pandemic platform were karnataka and telangana just to give you an idea of scale and the complexity we were going to delve into karnataka state is as large as france or even united kingdom and telangana is as large as the country of canada each participant was supposed to bring in their a game their a players their best technology solution the portfolio of products only then could we build this complex a problem what we needed like i said before is technology to be the backbone of this solution this pandemic solution but i must also tell you that not only did this solution needed a technology to be the backbone but needed this technology to have a soul because none of the participants were going to benefit by contribution they needed to bring their best technologies and they needed to provide the best portfolio components to build the solution for the need of the society for the population good this platform was built by multiple companies none of them own it this platform is open source people can add on top of it it's built with open apis it's cloud-based and it is for population scale so we have looked at implementing this two uh in into the two states that i talked about and and it is today still being used and will continue to be used and perhaps look at leveraging them for other states other large states that we have so in order to talk about know more about the technology solutions fractal was a key partner that you know was building the algorithms required building the solutions to address the needs we have for the government for health institutes as well as for the for the patients i would like to invite a shri kant to share the details on the solution that was built so she kant over to you my name is srikant velamkandi i'm a co-founder and group chief executive of fractal an ai firm for the next 15 minutes i'm going to describe the work we did with the governments of telangana and mumbai to manage the pandemic of covert 19. fractal's leadership team
was in london on february 25th and 26th for an ai conference followed by a company-wide town hall when it dawned upon us that kovac 19 crisis is getting worse and spreading globally we immediately imposed some travel restrictions and as soon as i got back to mumbai i decided to write to some state and local authorities telling them about the value of data and how analytics and ai could prevent the spread of the disease and help state and local authorities manage the pandemic better mumbai is a city of 20 million people a quarter of those 20 million commute by local trains every day the first few cases in mumbai were detected in early march and on 21st march i got a call from mr praveen pradesh the municipal commissioner of mumbai he wanted some help within a few minutes of that call i decided to call rahul sharma ceo of aws india and within a few hours of that we were on a long webex conference call with mumbai police as well as the mumbai municipal authority called bmc they needed help and the first thing they were looking to do was to track the number of arrivals in mumbai over the next over the last one month they want to look at everyone who had arrived in mumbai beginning beginning from march 1st and see where they were what kind of symptoms they had what were what were their activity levels and use that to to predict the spread of the disease soon the problem kept changing the problem immediately was after that was how do we model the spread of the disease then a few days later the problem became how do you manage healthcare capacity and then it was about how do we get patients into hospitals in as short a time as possible and eventually it was about how do we get industries reopened again so the problem the nature of the problem evolved every single day over the over the six months that we were engaged with the mumbai municipal corporation i'm not at liberty to tell you all the data sources that we used in solving this problem but suffice to say that the data was very complex and and it came from different sources and it was incredibly difficult to analyze and manage manipulate i wasn't very confident that we could put together any solution given the complexity of the data and given that the problem itself was changing on a daily basis but sure enough with the help of aws and our wonderful team who we called as fractal corona warriors we put together the solution within four days we had the first set of dashboards and decision cockpits ready for the local authorities by 25th of march coincidentally on 24th india declared a 21 days strict lockdown and it started on 21st 25th leaving the citizens of mumbai only four hours to prepare as we went on building the solution it became clear that there were four vectors of response first was to predict the spread of the disease and manage the spread of the disease it meant we had to build mathematical models at a zone level and at a ward level mumbai has six zones and 24 wards we needed to see how the disease will spread at each zone and ward level it meant we had to create a new metric called rt or real-time transmission indicator as opposed to r which is usually used in epidemiological models this helped us in finding out not just looking at the spread at a zone and ward level but how it was spreading on a real-time basis the next thing was to use this data and these estimates to predict healthcare capacity within the city of mumbai and soon it was telangana and karnataka the idea was to predict how many hospital beds would be required how many icos would ico bets will be required what is the need for healthcare workers ppes and even ventilators after that the other very important dimension of response was about making sure citizens understand the importance of this disease and importance of wearing masks and taking several precautions so that they could reduce the spread of the disease so we had to make sure that the information that was available to the local authorities was also a lot of that or most of it was available to the citizens and there were enough guidance provided to them and finally once this once india started unlocking after a few stages of lockdown we wanted to make sure that as industries opened we could look at the population that was traveling from their home location to the work location and what impact that that may have on the spread of the disease it meant that we could prioritize what kind of industries should open first and what next these were the vectors of response fractal has a very sophisticated data to decision process and we believe that it's not just about data science and machine learning but it's about design combined with data that really produces great results so fractal put together this process to not just design and frame and reframe the problem but to ingest all the data organize it process it and then design it for consumption and decision making that came in very handy in managing the pandemic we could spread we could look at the spread of the disease at a very hyper local level provide and provide citizens as well as the local authorities with decision-making tools so that they could they could inform the public as well as manage the disease in a better way we could also look at the spread of the disease at a very very micro level at a zone level at a ward level and and use that to make several decisions like create containment zones and so on they could also help in looking at the healthcare capacity and predict the healthcare capacity and make sure that no no at no point during the entire crisis there was complete shortages shortage of beds there was a time in in mumbai in may and june where mumbai was running with two or three hospital beds left in the entire city but then thanks to some of the work we did we could actually expand some healthcare capacity and manage and make sure that people didn't die needlessly once we started reopening the industry we started looking at how are people commuting to work and what that commute means do they bring the disease from where they're commuting from and do they actually take back the disease to their to their homes at the end of the day and what that would mean in terms of the spread of the disease i remember this extremely well in the middle of this pandemic once we had enough fatalities in mumbai unfortunately more than 6000 fatalities we decided to analyze all the data of the people who had who had passed away and seen and see what the trends underlying that and what are the risks of comorbidities and we held this two hour conference call with the top 10 doctors of mumbai who were charged with managing the pandemic overall and we could walk them through what the data said about the co-morbidities for example age what is the what was the correlation of age with increased mortality risk and some co-morbidities like hypertension tuberculosis and diabetes this immediately resulted in mumbai making several decisions about about senior citizens being quarantined and other kinds of decisions around treatment which also reduce the overall fatalities the aws intel fractal covert 19 platform provided us with advantages of agility flexibility and scalability we could build data pipelines quickly thanks to tools such as aws athena and aws glue to speed up data ingestion and preparation glue etl enabled us to integrate complex apache spark transformations without worrying about things like cluster resource management and high availability aws athena's serverless query service enabled us to query stored data on aws s3 buckets without the need for dedicated servers athena's seamless integration with tableau was critical as well and the aws team was always there to help and really agile given the rapidly evolving nature of our client needs flexibility was critical we could quickly leverage various sas tools such as map box and power bi depending on the need and from a scalability standpoint we were adding data sources every few days but we didn't have to worry about capacity constraints such as inherent because of the inherent scalability of the platform yes this pandemic is far from over but i feel great that aws intel fractal partnership could make a real difference to the city of mumbai and the state of telangana thank you
2021-02-08