CogniABle - Enabling Special needs lives the data way

Current studies suggest 1.6% ( 5 million) of children of 1-14 years are diagnosed with Autism. With limited clinicians available in our country, many children remain undiagnosed or diagnosed late.

Screening Service

A non-expert can use our screening service at 25% of the cost from any remote location by submitting a video using on internet. The video is analyzed by our proprietary and patented algorithms that measure behavioral and clinical landmarks of the children as young as 2 and makes screening outcome.

Intervention Support

Early intervention services can results in an increase of 10-15 IQ Points, however those services are not accessible and affordable. The access and affordability is brought by our online treatment service model that empowers parents and schools with the availability of an integrated assessment and intervention services available remotely at 25% of the cost. The treatment plan is behavioural bases and non-invasive.

1. Infrastructure and Clinicians supply demand Gap: In developing countries such as India, there exists a huge demand and supply gap for skilled clinicians (1 psychologist over 13000 individuals) and infrastructure facilities.
2. Delayed Screening of Autism happens in many cases in Tier 2 towns and villages leading to delayed start of early intervention services. 
3. Behavioral treatment for Autism is unaffordable as it costs more than $4300 per year (In India), an amount affordable by many without any insurance support. 
4. Majority of cases of Autism are found in Non-metro areas with rare adequate health-care facilities , clinicians and inclusive schools.
5. Poor treatment customization: Conventional treatment models offered are not customised as per the needs of the child and regular customization based on progress of the child in multiple associated factors. 

Our year 1 target market is to take our services through leading developmental pediatricians and schools in India. We plan to take our  screening and treatment services to the families through them. 

1. Through these leading developmental pediatricians in India we plan to reach 500 families in year 1 who visit clinics for regular therapies in pediatric clinics. The service positioning will be improved quality of care, access and reduced dropouts at clinics. From the same distribution channel we will reach additional 600-700 families who are located remotely and visit pediatricians in their clinic locations for quarterly reviews. Our service offering positioned across affordability, access and quality guides parents in the treatment journey. 
2. In addition we plan to work with 50 schools who are practicing inclusive education and assist them in capacity building through technology using our digital platform to start with our treatment platform immediately and integrate with screening platform in 12 months time. 

Our Solution to the underlying problem is offered through two digital platforms, given as:

1. Early Screening- Our early screening model is designed to overcome the challenges of misdiagnosis/underdiagnosis and early identification of on-risk cases of children on Autism. It uses AI-based algorithms that evaluate unstructured data gathered mainly from recorded video sessions to arrive at the screening result. It identifies gross motor and other hyperactivity patterns, and states the screening outcome with no waiting time. The parent/caregiver just has to record the video session of the child as per the guided script offered by the system. The uploaded video on the system is analyzed by the predefined algorithms to determine screening outcomes. The ai-based solution will be made available at the click of a button on mobile at just 25% of the current market cost.
2. Behavioral Treatment- The treatment model conducts the overall assessment of the child using structured and unstructured data reported by the caregiver of the child. The data is then analyzed by the predicting algorithm and designs the roadmap for the child. The roadmap will be target-based covering the overall development of the child which will help the child to reach the excellence point at an accelerated rate. The interface also provides the provision to track the progress of the child. The whole treatment pedagogy can be accessed remotely at just 35% of the current market cost. This will help parents overcome the financial and emotional constraints attached to the disorder.

Four main pillar of our innovative solution are

1. Used by Non-experts : The solution can be used by non-experts such as parents or caregivers in natural environment of child by simply playing and recording a video. The video is analysed by an algorithm with the same level of precision as that of clinician to make screening outcomes. This improves scalability and adoption in Tier 2 towns and villages. 
2. Affordable- Available for a price of $10 our screening service makes the solution affordable for the masses. Our behavioural intervention services priced at $2 a day makes evidence based intervention available to families having children diagnosed on Neuro-developmental disorder.
3. Remotely - Available remotely the solution can be used by any user who has access to internet
4. Application in other mental health areas - Our technology solution find multiple applications not only in child care but also in Geriatric care in both diagnostic and prescriptive applications.

We are currently using human action recognition framework to perform diagnostic and prescriptive solutions for children on Autism. This means an algorithm analyses a video and identifies key clinical landmarks with no human support. 

However action recognition task from video data set requires large amount of annotated data sets. We use a novel and patented technique to train a deep learning model to classify an action recognition task with minimum training data requirement by injecting guided noise in training data set. The model identifies and quantify fine motor, gross motor and complex actions of children with more than 94% accuracy. This allows an action recognition task to be identified with minimum downtime and training data.

In addition our treatment platform is developed using Python framework with Bigdata capability that allows machine learning models to analyze the data and customize the treatment plan for the child using child specific progress and other environmental and behavioral parameters. The algorithms customizes the plan for the child in the area of language , communication, social-skills and so on based on longitudinal progress data. 

We have conducted pilots for both our screening and treatment services. 

1. Screening platform: Proof of concept is in developed where machine-learning models were trained on expert annotated videos (6000) set of 37 children recorded over a three months period. The outcome is a classifier that can measure gross motor, fine motor and complex motor action skills of a child in various developmental areas. The prototype is patented as it allows developing a machine learning model to be developed with only 30% of the training data. The prototype is funded under prestigious Government of India grant (BIRAC) to develop Artificial Intelligence based digital platform to screen children on Autism. We have also been awarded prizes by Tata Trust, Lockheed Martin and Western Digital based on concept and real development development of our digital platforms. We have submitted the outcome of the experiments in a leading peer review journal that is under review stage.
2. Treatment service: We are undergoing Beta testing of our product with more than 200 families across 10 clinics in India and USA. Initial results have shown promising results where treatment services can be used by a non-expert at fraction of a cost with proven outcomes and treatment customization by Machine learning models.We are close to sign our work contract with leading hospital chain in India. The curriculum used in our treatment service is evidence based and licensed from US service providers with an agreement that it will be translated in local language. 

Currently, our treatment model has been deployed in 10 clinics across India and the US and currently 230 children are using the solution. The deployment of the model has generated enough insights for example making user interface intuitive and user friendly, making teaching pedagogy available in multiple languages this includes lesson plan, training videos and so on. 

In year 1 we plan to reach 1000 families and children in India (through pediatric doctor)  and atleast 50 clinics in the USA. 

Our innovation can impact 90 million lives in India and many more globally in first 5 years. The impact can be computed both on economical and social parameters. 

Affordable early screening services that can be run by non-expert , remotely can help identify children on neurodevelopmental delays. With 50K children born in India everyday we can touch 18 million new lives every year through our screening service.  

A conservative estimate suggest that in year 5 we plan to reach atleast 20000 families in India and in other countries on our digital screening and treatment platform. 

Our innovation can impact 90 million lives in India and many more globally in first 5 years. The technology eventually will lead to sectoral consolidation improving affordability, accessibility and quality.

Digital Screening Platform 

1. With 50K children born in India everyday we can touch 18 million new lives every year through our screening service. . 
2. The screening platform is under development with partners from leading government hospitals in India and universities and will go to market in year 2020-21. 

Digital Treatment platform

Autism is a lifelong condition that affects millions of children in India. 

1. In low resource country setting such as that in India augment skill and infrastructure deficits through capacity building with technology assistance. 
2. Fast-tracking adoption of technology by training sessions (online and in-person) and community engagement model headed by hospitals , not-for-profits and schools. 
3. We are looking to reach following targets 
   1. Enrolling 20 clinics in India and USA and bringing 1000 families to use our product (2019)
   2. Signing up with a leading hospital chain in India to provide digital solution to non-experts remotely (August 2019) 
   3. Developing machine learning models on digital platforms to reduce dependency on clinicians and allow families to start behavioral treatment remotely. 

The main three challenges that we foresee to come over are:

1. Health and Technological Connect: Bringing up the competency of AI-driven model with that of skilled clinician for accurate diagnostic results requires a large volume annotated video-data. The data should be based on wide range of children of different age, gender, geographical area, and skill sets. This requires a formulation of multiple partnerships with universities and NGOs to swift the early scaling. This also necessitates a blind site test for the tool validation and thus requires diversified data to ensure qualitative results.
2. Solution Adoption beyond Tier-1: An encouraging amount of potential customers reside in non-metro areas. Adoption of the self-paced/ customized treatment plan requires a basic understanding of technology and navigation of mobile application on a smartphone.
3. Revenue Model for customers in villages: The current revenue model is subscription based which requires monthly or annual payment to access behavioral treatment plan. This means the potential customer have access to means of digital payment through wallets or other means. 
4. Development of machine learning models on established clinical landmarks for classification and treatment customization objective. 

1. To overcome the need of large amount of annotated data we have built a novel patented technique that allows machine learning models to be build with 10 percent to 25 percent of the training data. This significantly reduces time for development of any human action recognition problem by 60 percent.
2. To scale our treatment services through a digital platform in India and internationally, our plan is to translate our service offerings in multiple languages. This includes lesson plan , videos and other resources that can be reused across population with similar assessment outcomes. 
3. We are talking to few representatives of the government to include our services by insurance schemes such as Ayushman Bharat. This may need additional time, however with focus on government to use digital assets, this is surely a way to go.  
4. We have partnered with leading educational institutions in India such as Indian Institute of technology Delhi (IIT) and few government and private hospitals to get best inputs and advise. 

Not applicable

We are a 19 member team this includes 3 founders. Our team  consist of 10 psychologist and 3 machine learning experts. We also have a HR lead and 2 member team with expertise in Django framework to develop web applications. 

Our team is balanced with right mix of individuals from management, technical and clinical backgrounds. We complement each other with expertise requirement to develop, validate and scale CogniABle across India and Internationally.

Partner Details

Manu Kohli is the Chief Executive Officer of the company. He has 16 years of technology consulting experience in multiple geographies. He has authored multiple books and paper in the area of Machine learning. He is a PhD candidate at IIT-Delhi with a focus on developing affordable screening and treatment solutions for neurodevelopmental disorders such as Autism. 

Dr. Prathosh AP is an assistant professor at IIT-Delhi and comes with 6 years of experience.He holds multiple patents and publications in the area of computer vision. He is central person who has developed various innovations and helped company file patents on them 

Dr. Josh Pritchard is a Board Certified Behavior Analyst (BCBA) and a professor at University of Southern Illinois, carbondale and comes with 18 years of experience. He runs various Autism clinics in United States and brings clinical expertise  that is translated back in our product development. 

Swati Kohli is a clinical Director of the Company and comes with 18 years of experience working with special needs children. She supervises in-house team of 12 psychologists and work closely in development of our Artificial Intelligence platform.

Diptiman is a marketing lead who comes with 18 years of business development and marketing experience in pharmaceutical sector. 

We are partnering with IIT-Delhi, a leading engineering institution to build AI capabilities. For developing clinical expertise we are partnering with few leading hospitals of India.

One of our partner and cofounder (Josh Pritchard) runs multiple Autism clinics and that allows us to bring back best practices of Autism treatment in India on our digital platform.

Our business model was chosen after conducting primary research with our customers and end-users both. Distribution Model ( relevant for both Screening & Treatment ) of our innovation is through developmental/Neuro pediatricians. 

Screening Service: The business model for our screening service is pay-per-use model where parents / caregivers can simply record a video of the child and submit to a cloud platform. The deep learning model of our service does the computation and shares the result with no waiting time. The model allows screening to be performed for families with no commute , higher accuracy and at 30% of the price. 

Treatment Service: Our treatment service is available on monthly subscription model at 1/3rd of the price. The platform enables parents to be primary caregivers to deliver high quality behavior treatment. These services are available to parents/ caregivers in the comfort of their locations, using teaching pedagogy with detailed lesson plans, video pedagogy and 4 hours of personalized consultation sessions with our experts. 

1. Unit Economics ( Relevant for Treatment ) 
   • Revenue -5000 INR/user/Month 
   • Cost - Cloud -300 INR
   • Staff Cost ( 25000 INR Per month)- Per unit customer 25000/20=1400 ** ( This will potentially reduce in coming years using Machine Learning and AI to 20%) 
   • Rental - 300
   • Overheads - 300
   • Variable Costs ( Support function salaries , Product maintenance, )-500
   • Marketing & Travel - 500
   • Misc-500

Margins= 5000-3800=1200 

In 2018 based on our patented prototypes government of India funded the company under BIRAC BIG scheme. A grant of $70k was issued to develop Artificial Intelligence based screening solution for Autism to our company. We have also received various prizes from contest that are sponsored by organizations such as Startup-India, Western Digital, Tata Trust and Lockheed Martin. We are using current set of prize and grant money to develop  screening and treatment platform.

We are also negotiating with few funds to invest $1 Million in our company. 

The funds that we plan to raise will be used for product development, reaching wider audiences and scaling our business. Our revenue model even though affordable, is cash positive from unit economics perspective. We will be generating revenue from product sales to ensure self-sustenance and covering our variable costs such as salaries, rentals and so on. 

We are applying to solve mainly because of the following reasons:

1. Technical Expertise- To seek technical expertise that MIT will render on the selection of the proposal to make the product thrive in the market.
2. Scaling- Advice on strategies to scale up in non-metro areas and adoption of revenue collection channels in such areas.
3. Financial Assistance- we are looking to fulfill our financial requirements associated with product development and scaling the business beyond national boundaries.
4. Building relevant connections- To develop relevant connections with industry experts who can advise on innovation and expansion in all directions.

We will like to partner with following organizations

1. Leading hospitals to guide our team on clinical parameters important for developing Autism Screening and treatment solutions. 
2. Clinics to pilot our solution with established feedback and engagement model. 
3. Leading Cloud service providers such as Google, Microsoft or Amazon to assist us in product development and development of machine learning models. 
4. Partner organizations that can help us scale our services in other geographies. 

Disruptive behaviors are major cause that inhibits learning of a child diagnosed on Autism. Understanding function of behavior is a challenging task that requires presence of skilled clinician who observes the child and records functions of behaviors, observational data with detailed antecedent and consequence analysis. A detailed behavior plan is then prepared and implemented for the child involving family members, school teachers and clinicians. 

We propose to use Deep learning models to perform behavioral measurement for few disruptive behaviors such as headbanging and handflapping. Our patented algorithms that performs human action recognition will be used on this task. We will implementing our algorithms on recorded video streams of children who are engaging in this behavior in a natural environment. The model will uses methods of action recognition, segmentation , localization and stimulus response measurements on complex self-stimulatory behaviors. The algorithm will suggest statistical measures on disruptive behaviors ( frequency, IRT, etc) to draft behavior plans and analyze functions of behaviors such as antecedents and consequences. 

Our company is developing screening solution for Autism where a non-expert can engage with a child in a playway approach and record a video. This video is analysed by Machine learning models to make Autism Screening predictions. 

We propose to train 100 Anganwadi workers ( community workers) in a pilot who are engaging with family at a grass-root level. The workers foster participant involvement by creating awareness and engagement with families. Majority of these workers are females who interact with other females in community and monitor immunization schedules for the child till the age of 5 years. These workers will be empowered and trained to teach mothers on how to play with a child based on a script for 20-25 minutes so that the session can be video recorded and submitted on a cloud platform so that it can be analysed by an Artificial Intelligence model.

This will ensure that lack of infrastructure and clinicians does not devoid a family to identify a child on Autism early. The will of a mother to learn new skills and with assistance of technology a child can be screened of neurodevelopmental delays such as Autism remotely, by non-expert and at affordable cost. 

Our current digital platform is hosted on a cloud. We plan to use longitudinal behavioural treatment data of 500 families with informed consent to build Machine learning model that will customize the plan as per the need of the child. That allows caregivers to optimize the teaching pedagogy and match it to optimum potential of the child in areas such as social skills, language, life skills and so on.

Our biggest challenge in development and generalization of Machine learning model is the availability of  data with wide representation and variety. We will engage with families who have children in age-group of 2-5 years and have formal diagnosis of Autism. A formal ethical committee with a local hospital in India will be formed to ensure necessary standards for data collection are adhered and followed. Clear inclusion and exclusion criteria for the study will be defined with an informed statement on how and in what cases the data will be used. 

The data will be stored on the cloud with access limited to 2 people as per our organization roles and responsibilities policy. The physical location where all the forms, signed informed consents and videos will be stored will be inaccessible to other team members in the organization.