vLife is Virtusa’s own healthcare and life sciences marketplace for anyone who deals with data and data science. Through the self-service portal, payers, providers, biopharmas, and medical device companies can access industry-based solutions built on journey maps.
The solutions include pre-trained machine learning (ML) models, jump-start snippets of code, tools, and accelerators specific to the healthcare and life sciences industries. Built on AWS, vLife provides a comprehensive set of solutions on a HIPAA-compliant data lake with multiple sources, prebuilt APIs, AI, and ML models.
It offers niche tools that not only leverage the latest technology but also address the domain needs. It is an AI/ML-based applications marketplace. vLife contains over a billion rows of clinical, claims, genomic, and IBM™ MarketScan™ research data. In addition, vLife also contains simulated data for full U.S. population for 34 disease conditions.
Using vLife, customers can get access to and deploy code based on their unique business challenges. The code can act as accelerators for prediction models, and visualizations to create a convenient shopping cart checkout, recommend similar products, and establish buying patterns based on purchase history. With close to 1,000 AI artifacts, customers can build their own ML models to leverage data based on their priorities and budget. The extensive collaboration is made possible through GitHub supporting projects.
While there have been numerous technology-driven advancements in the healthcare and the life sciences industry (thanks to the AI revolution), our goal is to functionally slice the business problems into service lines, platforms, LOBs, and use cases. These use cases are divided horizontally across domain-centric themes, and vLife recommended technology offerings like Faster AI, Computer Vision, Synthetic Data Generator, and IaaS.
Based on the belief that it’s imperative to use technological advancements to fuel our research towards value-based solutions, vLife provides a holistic approach, covering each service line and domain-driven platform.
The functional goals are:
We developed each solution on vLife with a customer-centric, research-driven approach and separated the offerings into specific service lines. Our solution-specific platforms include:
In addition to this, we also have the vLife Offerings Platform with Faster AI, Computer Vision, Synthetic Data, and Innovation as a Service (IaaS), which horizontally cuts across all these solutions. We have a separate section on the vLife Offerings.
We validate the success of a model against the domain-driven use cases daily by comparing it against the ML problem (framed) and the system level resources used to support the phases of ML workloads. Our end-to-end solution building life cycle contains the following set of activities with automation applied at every stage:
We use Amazon S3 as a data lake to hold large data volumes such as electronic medical records (EMRs), claims, etc. An Amazon SageMaker notebook to run against the P3 instance types and G4 EC2 instances. Data cleansing, processing, discovery, and feature engineering get completed in both the SageMaker training instance and G4 EC2 instances. We also use Amazon SageMaker to train custom models using the transformed data, leveraging the distributed training capability, and creating an Auto Scaling API endpoint. The API endpoint makes batches and real-time inferences.
vNet is an enterprise-AI application for building faster, smarter, and better ML models to accelerate digital transformation. vNet has three components, data, AI, and business, that are further broken down into platforms.
Artificial intelligence (AI)
vNet is built for domain experts, data scientists, and data engineers to:
The five key offerings of vNet within vLife to accelerate the technology-driven innovations are:
Platform AI provides a platform that combines all AI-driven solutions in the form of data, NLP, visualization, and insights designed to empower customers across any domain to explore, train, experiment, and identify domain-driven solutions for conclusive decision-making. It is a framework that functions more efficiently and intelligently than traditional ones to help clients scale from experimentation with a domain-driven, structured approach.
Faster AI is an intuitive AI tool where you can easily build, train, and deploy ML models, all in one place, without writing a single line of code. This platform enables users with no AI coding skills to optimize day-to-day operations, allowing developers to rely on technical purity and solve business issues. Faster AI empowers teams to operationalize AI and drive business value.
Leveraging the tools from SAIL, we’ve optimized the model to obtain a noise-free and unbiased dataset.
Computer Vision is a no-code/low-code image analytics platform that helps optimize performance to increase efficiency and decrease operational costs. With the ability to upload your videos or images, you can carry out the necessary pre-processing steps, train AI/ML models, and then start using them. Development of the computer vision models is completely codeless. Everything on the platform is an intuitive, visual task, resulting in an environment that takes only a few minutes to learn.Thus, it becomes a powerful tool for beginners to get started quickly and AI experts to develop faster and leverage custom code with a fully managed infrastructure for computer vision.
Synthetic Data Generator is a platform that uses algorithms to create artificial data by mirroring the statistical properties of the original data without revealing any information regarding real people. Using the platform enables business users to maintain privacy, successfully carry out product testing, and train ML algorithms. It helps businesses avoid choosing between data privacy and data utility while selecting a privacy-enhancing technology.
Our AI-powered synthetic data solution takes your original data and transforms it into privacy-compliant synthetic copies. Data in the form of tabular datasets or images can flow freely across your business and partnerships while providing maximum data security. By retaining the original data with synthetic copies, we empower engineers, data scientists, analysts, and product owners to make decisions that matter faster — without exposing your sensitive data.
The platform makes building datasets faster and better with data quality profiling, labeling, and synthetic data generation. It helps anyone who struggles with gaining access to sensitive data or uses poor quality data while building and deploying scalable AI solutions. Our platform solves these data pain points with synthetic data and tools that improve data quality with automation.
IaaS is a platform that intuitively connects every step of the domain journey and hosts several solutions. These solutions help customers efficiently improve the cognitive system of the domain systems across various AI-driven innovations. The ability to capture new innovations from the knowledge pool and make them accessible to the corporate world is what we call Innovation as a Service.
Our partnership with Stanford Artificial Intelligence Lab (SAIL) and the use of Generative Adversarial Networks (GANs) differentiate vNet from other solutions.
vNet allows users to perform the following operations:
The benefits of using vNet via vLife are the ability to:
With over 1,000 ML models and AI artifacts available on vLife, you can build your own ML models to leverage data based on your organizational priorities and project budget.
Training customized ML models will remain a challenge for data scientists if they don’t have enough data available. To address this, we’ve created vNet. The solution, available on vLife, gives organizations access to high-fidelity synthetic tabular and image data generated by vLife applications.
Using vNet, you can get access to:
We recognize the critical importance of collaboration in advancing AI technologies to create intelligent solutions. These solutions have the power to make lives better, businesses more efficient and help them evolve as thought leaders in the healthcare and life sciences space. We understand that not all solutions can be built and hosted in a single place as one “solution for all”.
vNet will soon be accessible for all Virtusa customers and partners via vLife healthcare and life sciences marketplace in a UI-based application within vLife.
Our goal is to develop cutting-edge assets, guided by academic research and partnerships, that address industry pain points. Through a rich marketplace experience, we can inspire everyone in the healthcare and life sciences industries to accelerate AI-driven medical research with data and innovation.
We designed vLife to give a smooth user experience to its trusted users. The User Management Engine of vLife was developed using AWS Cognito to protect the integrity of our customers and is 100% safe and secure. The data is securely stored in AWS Cloud and is under multiple firewalls and security layers for protection.
The User Management Engine is an end-to-end workflow that stores, maps, protects, and classifies the user’s information using AWS Cognito and other AWS services. It guarantees 100% security over user information. Credentials used to access vLife are converted to secure access keys and are encrypted multiple times. Once the user registers/logs in to the vLife portal, the credentials are verified in the backend incognito. We verify the user’s email with a one-time password (OTP) for extra safety, which remains valid for 24 hours. The user is then mapped to their subscription plan and can access the vLife offerings. These credentials remain unique for every user, thus protecting integrity.
For every subscription plan, there’s a primary user representing the organization. The primary user has admin rights to allow access to and deny access from additional users for the organization based on the subscription plan.
While registering to the portal, the user needs to verify their email, generate a password, and provide the company’s name. After successful registration and approval from a vLife admin, the user can then directly log in to the portal using their username and a password.
Synthetic data helps solve the most common data problems by either producing artificial data from scratch or using advanced augmentation techniques to produce novel and diverse training examples. We use the following tools to generate synthetic data:
Capturing innovation from crowdsourcing is now streamlined enough to become a part of any company’s survival kit. Crowdsourcing relies on large, unstructured groups of people to create a single, structured outcome. Businesses want to harness the crowd to expedite innovation and accelerate product and technology development. Capturing new innovations from the knowledge pool and making them accessible to the corporate world is essentially IaaS.
Virtusa joined the Stanford Artificial Intelligence Laboratory (SAIL) as an Affiliate Member through its vLife initiative. The collaboration brings together ‘Virtusa’s deep digital engineering expertise and prestigious researchers from ‘the Stanford Innovation Lab to develop and advance AI technology. With SAIL collaboration, vLife will focus on using AI to help patients live longer and improve the profitability of products used across the healthcare and life science industries by extending open-source research projects for AI.
The SAIL Applications have also been integrated into the Faster AI platform of vNet that users can use to create noise-free and unbiased datasets. Virtusa and Stanford Artificial Intelligence Laboratory (SAIL) developed an efficient approach to help find and mitigate bias from our ML models. Pre-processing and post-processing of both datasets and models create our systematic approach.
vLife is designed based on the AWS Well-Architected Framework embracing the architectural best practices for designing and operating reliable, secure, efficient, and cost-effective systems in the cloud. The architecture is consistently measured for the best practices focusing on how to design, deploy, and architect our ML solutions in the cloud.
The design as such enables developers and data scientists to quickly and easily build, train, and deploy ML models at any scale using AWS SageMaker.
In vLife, the ML operations are driven using open source ML frameworks, such as TensorFlow, PyTorch, and Apache MXNet.
The Deep Learning AMI and Deep Learning Containers in this level have multiple ML frameworks preinstalled that are optimized for performance and serve as the core engine for Faster AI. The optimization makes these models always ready to launch on powerful, ML-optimized compute infratructures, such as Amazon EC2 G4 and P3 instances, which provide faster speed, cost optimization, and efficiency to ML workloads.
At vLife, we follow the Seven Steps to Excellence for MLOps:
Federated learning is an ML technique that trains an algorithm across multiple decentralized edge devices or servers holding local data samples without exchanging
them. In vLife, we aim to implement federated learning when training the distributed data. Differential Privacy is maintained, and Deep Neural Networks’ accuracy of
case-sensitive medical data is not compromised.