Swaasa AI Platform for respiratory healthcare
Global burden on respiratory disease (as per 2011 census)
550 million patients with respiratory diseases.
1 in 8 deaths attributed to respiratory diseases.
7 million deaths cause by respiratory diseases.
The challenge we seek to address is the “The inability to screen, diagnose
and monitor lung health at scale, in real-time.
Screening for respiratory diseases is an unmet need due to limited human expertise and lab facilities. It is not possible to do pulmonary function tests at primary care level centers. Any technology / tool which can facilitate screening and identify whether the problem is attributed to airway or lung parenchyma/pleura will be a highly useful value proposition for the appropriate next intervention.
Currently the products used for screening are Stethoscope, X-Rays and Spirometry, all of which require qualified trained personnel and special equipment too.
Challenges in current process
1. Complex Assessments - Tests are lab-driven and requires trained staff and visits to hospitals
2. Expensive - Each test incurs a cost of
approx. USD 1000 for individual, insurance or health systems
3. Reactive - Action only on realizing worsened respiratory symptoms and heightened risk
4. Limited self care - Extremely dependent on doctors
plus unavailability of doctors in remote locations
Swaasa AI Platform, a patented ML technology that brings the functionality of spirometry without the constraints of time, space, specialized equipment and trained professionals. The cloud based Swaasa AI platform analyzes cough sounds along with a subset of symptoms from SGRQ (St George's Respiratory Questionnaire) and provides an assessment.This core ability allows the solution to be rapidly deployed at scale to screen, diagnose and monitor millions of people suffering from the Big Five respiratory diseases.
Swaasa AI platform is mainly targeted as a screening tool and diagnostic aid for respiratory diseases in low resource settings. The platform provides easy and cost-effective tests to reach larger populations at a much higher frequency of testing.
The Swaasa AI Platform
1. does not require trained professionals
2. does not require any specific hardware - can work on smartphone, tablet or laptop
3. no other consumables needed.
4. can be enhanced to be used with landlines / feature phones via IVR.
5. requires very low bandwidth, no bluetooth required.
6. low battery powr consumption - If using a smartphone, then atleast/approx 360 assessments can be done without any battery recharge of the smartphone.
Preliminary work carried out:
1. Established clinical evidence of the technology through (a) publications https://www.sciencedirect.com/science/article/pii/S2352914819304071 (b) Clinical Evaluation Report submitted for ISO13485 certification.
2. Till date completed 5 clinical validations covering more than 2500 subjects. The validations primarily focused on identifyin
(a) Underlying respiratory problem (yes/no)
(b) Pattern of the respiratory disease (normal/obstructive/restrictive/mixed)
(c) Severity of the pattern
(d) Likely presence of COVDI-19 (yes/no)
(e) Likely presence of PTB (yes/no)
The product variant with functionality (a), (b) and (c ) is called Swaasa AI Platform 1.0, which was validated in a clinical setting and also at a primary health setting. This version also has initial commercial deployments.
Further enhanced the product by adding an ability to detect the likely presence of disease conditions for COVID-19 and Pulmonary Tuberculosis. The corresponding product version is Swaasa AI Platform 1.2 and Swaasa AI Platform 1.4. Completed clinical validation for the same recently.
In addition to the above a large-scale validation on more than 3000 subjects (a) to detect likely presence of disease conditions Asthma, COPD, Bronchitis, Pneumonia and ILD (b) and to predict spirometry values based on cough sound characteristics is in progress at Christian Medical College, Vellore, India. Expected to complete by Jun 2022.
Swaasa assessments are
Contactless - Swaasa assessment involves no physical exposure of any kind / in any way
Realtime - The test is quick, and returns a result in a matter of seconds
Affordable - Less than $0.15 per assessment
Accessible - Anywhere available, anyone can operate
Transparent - Testing system provides reports and essential visibility into staff health
Continuous - The test can be taken multiple times a day to assess incremental changes
Scalable - The platform is proven for 1 Million assessments / day. Can be scalled to 500 Million.
Our target population is people living in remote and rural areas in resource limited LMIC settings.
Swaasa AI platform is mainly targeted as a screening tool and diagnostic aid for respiratory diseases in low resource settings. The platform provides easy and cost-effective tests to reach larger populations at a much higher frequency of testing.
The leadership team has right blend of business experience and product development experience. Total we bring more than 85 years of industry experience. The product development and QA team comes with very strong expertise from Machine Learning, Signal Processing and Verification and Validation domain. We are also well connected with other organizations who are helping us on market research, marketing collateral.
The founder had identified the opportunity based on a discussion with Dr. Anand Krishnan, MD, PhD. Professor, Centre for Community Medicine - All India Institute of Medical Sciences, New Delhi, India - "Screening for respiratory diseases is an unmet need as due to limited human expertise and lab facilities, it is not possible to do pulmonary function tests at primary care level centres. Any technology / tool which can facilitate screening and identify whether the problem is attributed to airway or lung parenchyma/pleura will be a highly useful value proposition for the appropriate next intervention". Swaasa tool developed by Salcit Technologies addresses this need.
Since then we are continuously interacting with
a) Leaders from Community Medicine. Interacted with Head of Department, Community Medicine Andhra Medical College, Kasturba Medical College, Christian Medical College. Spent time in visiting community health centers, primary health centers and sub centers. Understood how the current process works for screening respiratory problems. We also visited Chest hospital, Andhra Medical college to understand the screening diagnostic process for Pulmonary Tuberculosis.
From all these interactions it is very clear that today for screening respiratory problems the dependency is still on chest x-ray and spirometry. These are not available in many of these centers, in places where these are available there is too much dependency on skilled technicians / manpower.
Today at Rural Health Center, Visakhapatnam the medical officer makes a decision on referring a patient for further investigation based on symptoms information gathered including medical history from the patient. With the ongoing COVID-19 pandemic, the use of stethoscope for auscultation has come down drastically. And in many cases the medical officer / health-worker is not skilled to do auscultation. So many patients who visit the center are undiagnosed or underdiagnosed and they are not provided with any objective evidence on why they should go to a tertiary care hospital for further investigation.
At sub centers the condition is much worse, here the front line workers are not equipped with any devices / test facilities. They are looking for technologies like Swaasa which help them in doing lung health assessment by just using a smart phone. They can then immediately look at the result and take opinion from medical officer / doctor for the immediate next intervention. Today they are using questionnaire. This is very subjective assessment. We did a comparison on how Swaasa is improving in case finding by using questionnaire at one center and by using Swaasa at other center. We could see 30% increase in case finding with Swaasa which otherwise were getting missed out. Underdiagnosis a major issue.
b) Health service providers. Providers that are primarily providing health services in rural and remote areas. Some of these service providers have well connected online consultancy and related services based on the screening at remote health centers. Unfortunately, at these remote centers they dont have spirometry setup for screening lung problems. In cases where they are using handheld spirometry there are problems as the spirometry require an active participation from the patient and also requires a technician to guide for a proper test. Not every patient can perform / undergo / participate in spirometry. We could see an opportunity of integrating Swaasa for lung health assessment with their already existing screening tools for vitals and other health conditions. The output from Swaasa is readily available to the doctors to guide patients on next steps.
c) Doctors in private clinics in Tier-3 and Tier-2 cities.
The person in context is a physician who works in a private hospital as consultant and also owns his own private clinic in a Tier 3 city. These clinics are visited by people, who are suffering with seasonal changes; experiencing some symptoms like cold, cough, breathing difficulty from last few days and patients already undergoing treatments for some diseases. In most of the cases the visits are for non-emergency conditions. Once the patient interacts by providing basic information then the doctor asks few questions like how long he has cough, does he also has sputum, wheeze, throat pain and so on. Doctor also does Auscultation to see whether the chest is clear or not. Anything related to fever, pain, and cold, allergy the doctor can more or less identify the problem with confidence and does the required prescription. When it comes to respiratory problem identification, based on chest auscultation (if it is clear) and patient has cold he comes to conclusion that this may be related to Allergy. In some cases he asks to get a blood test done to know the Eosinophils. For Allergy he prescribes the drugs. Generally patient won’t come again if he comes then the problem is something else. If the patient says he/she suffering with breathing difficulty he records the history of the patient. Especially he goes deep to know whether the patient had Asthma, COPD or other respiratory diseases in the past. From chest auscultation he tries to understand whether there are presence of sounds like wheeze and crackles. If they are present he tries to understand whether they are local or generalized. If sounds are local and not very intense and if the patient has history he will suggest a medication like Inhaler with albuterol sulfate, inhaler duolin etc. If sounds are local and intense or if sounds are generalized he will request the patient to see Pulmonologist. In some cases he might request a Chest X-Ray or a PFT test, but not being a specialist in this domain mostly he prefers patient seeing the pulmonologist. The problems he face with auscultation are (a) he may not be able to catch sounds which are low intense / weak (b) not able to spend quality time in doing auscultation at all important points. The other problem is cough being an important symptom he just relies on patient information about frequency of cough, not skilled enough to find out possible reason based on cough sound pattern. Any test he suggests to patient is expensive (PFT is not easily affordable in Tier 3 cities and may not be there in many places). Here is where the doctor feels the need for better methods to identify the problem with confidence, use the indicative information as cross reference data and based on this information take a firm decision for patient to see a pulmonologist or not. He also feels any such method if that can be performed by Nurse Assistant by spending more time on patient that makes his job lot more easy and effective. Any such new method he would like to adopt provided it is lab free and cost effective, and easy to use by his staff. Furthermore he feels any such method must have a mechanism to relate the findings from multiple visits. This helps in knowing whether the condition is deteriorating or improving. He also believes he is saving cost to patient for cases where there is no need for PFT or X-ray, but his assessment (being inconclusive) prompted him to suggest the X-ray or PFT or both.
Overall we interacted with more than 50 physicians, more than 20 pulmonologists and with various service providers to understand the problem as part of primary market research. Based on this we defined the product usecases for (a) screening and triaging (b) treatment efficacy and disease management (c) predict exacerbations. Currently we are focusing product offerings for (a) and started conducting longitudinal studies to plan future enhancements for (b) and (c).
To illustrate an example the interaction with these people helped us to enhance Swaasa by having a functionality to do self assessment at home. Doctor can choose to send a SMS (text) link to patient. Patient can then open this link at home using a browser and can do cough recording using a smartphone. The result of this test is immediately available to the Doctor.
In addition, we are also working with Doctors who are playing the role of principal investigators for validating our product. We are getting very valid inputs from these doctors in refining our product functionality. All this helped us in defining the right intended use and indications for use. And in defining use case for integrating into the clinical flow. For example in every hospital the nurse will be doing the lung health assessment using Swaasa along with other vitals which are getting captured. When the patients goes to Doctors office, the doctor has all the information about the patient lungs, this helps in making an informed decision on the possible next step.
Similarly, we got input on how Swaasa can be used on Post COVID-19, COPD diagnosed patients as part of disease management. This is helping us in planning longitudinal study for the product.
- Identify, monitor, and reduce bias in healthcare systems, including in medical research and at the point of care
- Pilot
We are primarily looking for
1. Strategic advice from experts
2. Receive access to investments
3. To be a part of network of impact-minded leaders
4. Help us enhancing the impact we can create
5. Gain exposure
- Business model (e.g. product-market fit, strategy & development)
Swaasa AI platform is mainly targeted as a screening tool and diagnostic aid for respiratory diseases in low resource settings. The platform provides easy and cost-effective tests to reach larger populations at a much higher frequency of testing.
The Swaasa AI Platform
1. does not require trained professionals
2. does not require any specific hardware - can work on smartphone, tablet or laptop
3. no other consumables needed.
4. can be enhanced to be used with landlines / feature phones via IVR.
5. requires very low bandwidth, no bluetooth required.
6. capability to work on battery with lesser charging time - If using a smartphone, then 360 assessments can be done without any recharge.
Think of our tool as
1. Point of Care, Software as a Medical Device for instant evaluation of respiratory health
2. The equivalent of home monitoring for Blood Glucose and Blood Pressure - for respiratory health. Especially, for your Long Covid and Lung Fibrosis platforms, we may be a perfect complement.
We expect Swaasa to revolutionize the respiratory healthcare. Swaasa aims at providing respiratory healthcare unbounded by location, time, equipment, and technicians.
For the next one year the impact is
We are targeting 1 million end beneficiaries in 12 months through our B2B customers and partnerships with healthcare providers / healthcare organizations. This is primarily in low resource settings. Essentially our aim is to detect the reporiatory diseases including PTB early by doing screening in low resource settings. The accessibility and affordability will help in maintaining good health and well being.
Assuming each end user requires a minimum of 2 tests, the number of tests is 2 million.
On an average assuming close to 40 minutes savings in time for screening lung condition compared to existing methods and 100 USD savings from the test cost for 2 million tests we anticipate to save
13 million hours of time
200 million USD
For the next 5 years our target is to reach 100 million users from LMIC countries. This has an impact of saving
1. 130 million hours of time
2. 20 billion USD
The actual savings can be much higher than above as we did not factor the time it takes to get a report for the test. And importantly the cost incurred by the healthcare system due to delay in identifying the disease.
Using Swaasa can further can help in reducing the
Number of outpatient visits
Number of ER visits
Hospitalization
Repeat hospitalization
Our strategy to enable access to new markets is through channel partners / distributors. We have partnerships with companies like Stratmed, Charmhealth who are helping us in reaching hospitals and health service providers.
We are identifying partners like Bangladesh Grameen bank, SAMRC (South African Medical Research Council) to extend our reach to various other countries.
One aspect important for us is to have regulatory approvals in these countries.
We have ISO13485 Medical Device Certificate and on our path to get CE certificate. In India we currently have no objection certificate as our device comes under non regulated category, but by Aug 2022 the government is planning to regularize all devices include software as a medical device. We have all necessary validations, certifications for us get the approval. This will help us in getting regulatory approvals in countries like Bangladesh. The CE certificate will ensure access to other markets.
Our indicators for measuring progress are
1. Number of assessments / day. Our immediate target is to ensure that on an average 5000 assessments / day are getting done using Swaasa. This will help us in reaching 2 million assessments / year.
We will keep pushing our partners / customers and also acquire more customers in maintaining this number. We have a mechanism to track through a dashboard the status of the same.
2. From 5000 assessments / day we would like to move on to 0.5 million assessments / day over a period of next 5 years. This will help us in reaching 200 million assessments / year.
3. Number of dollars saved
4. Time saved
Respiratory diseases impose an immense worldwide health burden. Five of these diseases are among the most common causes of severe illness and death worldwide. Respiratory diseases make up five of the 30 most common causes of death. Respiratory diseases account for more than 10% of all disability-adjusted life-years(DALYs), a metric that estimates the amount of active and productive life lost due to a condition. Respiratory diseases are second only to cardiovascular diseases (including stroke).
Altogether, more than 1 billion people suffer from either acute or chronic respiratory conditions. The stark reality is that, each year, 4 million people die prematurely from chronic respiratory disease. A total of 9 million children under 5 years old die annually, and pneumonia is the world’s leading killer of these children.
Even more distressing is the enormous suffering that living with these illnesses cause. Those who are most disadvantaged suffer most from poor health. With this awareness, the United Nations (UN) created the Sustainable Development Goals (SDGs) in 2016 to raise living standards globally.
In May 2014, the World Health Assembly approved the End TB Strategy, a comprehensive set of principles and activities developed by WHO’s Global Tuberculosis Program over a 2-year period with broad global consultative input. Three main areas of activities (or “pillars”) are necessary to achieve these targets are identified as:
Pillar 1 Integrated patient care and prevention
Pillar 2 Bold policies and supportive systems
Pillar 3 Intensified research and innovation
While, preventing and combating respiratory disease is a highly cost-effective “best buy”, this falls outside of the purview of the current project for the most part. However, once disease occurs, the goal is to lessen its effects and cure it if possible. Reducing its effects is best accomplished by early detection, prompt diagnosis and early effective treatment. Specific targeted national (or regional) interventions should prioritize interventions that are effective, easy to implement and monitor, and that target persons most at risk. For example, in asthma, this could include a programme with early treatment with inhaled corticosteroids, reduction in exposure to tobacco smoke and appropriate education in the use of inhalers.
In summary, based on best evidence, there is an acute need for new ways to triage, screen and diagnose [detect] and monitor respiratory diseases at scale in resource poor settings. In other words, a simple easy to use low cost way to achieve these objectives which can be put in the hands of either the (remote) at risk populations and/or community health workers has tremendous value in this global fight against respiratory diseases. This project fills precisely that gap.
Below are Theory of Change pathways that this intervention would lead to.
Impact
- Contribute to achieving eliminate TB goals of WHO and Government of India
- Contribute to surveilling and monitoring in real time detection and spread of various respiratory diseases at the population level.
Outcomes
- Early detection of respiratory diseases (both NCDs and infectious diseases)
- Reduced deaths
- Reduced burden of families
- Reduced caseloads on health systems
Output
- Robust platform tested at scale
- Ability to handle numerous respiratory diseases
- Obtain regulatory approvals around the globe
Activity
To build the ability to triage, screen, diagnose and monitor lung health at scale, inexpensively and with minimal resources.
- Long term
- Deploy at scale
- In self-serve model
- In hands of Community Healthworkers
- Extend to pediatric patients
- Add additional disease condition
- Add additional biomarkers (breath sounds and speech)
- Deploy at scale
- Short term
- Train algorithm to detect likely presence of disease conditions (COPD, Asthma, ILD, Bronchitis, Pneumonia)
- Pilot deployment
Inputs
- Funding for furthering the development
- Access to multi-site (multi-national) replication of studies
- Government(s) support for deployment
Problem Statement & Assumptions
Sustainable Development Goal 3 (SDG) Target 3.3: By 2030, end the epidemics of AIDS, tuberculosis, malaria and neglected tropical diseases and combat hepatitis, water-borne diseases and other communicable diseases
- In 2015, 10.4 million people developed tuberculosis (TB) and 1.4 million people died from it. 3.6 million people with TB are missed by health systems every year
- Globally deaths due to coronavirus (COVID-19) pandemic are reaching 6 million
- Respiratory diseases contribute to over a quarter of NCD deaths globally each year
- Currently majority of the respiratory diseases go undiagnosed
Cough is the most common symptom of several respiratory diseases. Cough is a defense mechanism of the body which prevents respiratory tract from foreign materials inhaled accidentally or produced internally by infections. Changes in the character of the cough sound can reflect pathological situations in the lungs. Pathological situations arise due to some conditions like obstruction, restriction, and combined patterns.
We were able to identify a few conditions of importance like widened airway, narrowed airway, airway narrowing due to loss of elastic support, inflammation of lining of the bronchial tubes, air sacs filled with fluid, lung compliance and the interstitial becoming scarred and thickened. We collected cough sequences and annotated them based on these identified conditions, extracted features and did a thorough analysis in relating the feature characteristics (cough signature)to these different conditions.
Once the above evidence got established we used Machine Learning technology to train algorithms for these different disease conditions and further to predict based when a new instance is given as input.
Primarily our technology stack involves
1. Audio signal processing
2. Artificial intelligence and machine learning
3. Mobile and web applications
4. AWS cloud infrastructure
5. Dashboards for analytics and visualization
- A new technology
1. Established clinical evidence of the technology through (a) publications https://www.sciencedirect.com/science/article/pii/S2352914819304071 (b) Clinical Evaluation Report submitted for ISO13485 certification.
2. Till date completed 5 clinical validations covering more than 2500 subjects. The validations primarily focused on identifying
(a) Underlying respiratory problem (yes/no)
(b) Pattern of the respiratory disease (normal/obstructive/restrictive/mixed)
(c) Severity of the pattern
(d) Likely presence of COVDI-19 (yes/no)
(e) Likely presence of PTB (yes/no)
The product variant with functionality (a), (b) and (c ) is called Swaasa AI Platform 1.0, which was validated in a clinical setting and also at a primary health setting. The above publication covers the details of this.
Further enhanced the product by adding an ability to detect the likely presence of disease conditions for COVID-19 and PTB. The corresponding product version is Swaasa AI Platform 1.2. Completed clinical validation for the same recently. Publication submission for these are in progress.
- Artificial Intelligence / Machine Learning
- Software and Mobile Applications
- 3. Good Health and Well-being
- 11. Sustainable Cities and Communities
- India
- India
- For-profit, including B-Corp or similar models
Salcit as an organization is committed to promote diversity & inclusion.
The organization supports the development approaches that lead to gender equality and women's empowerment. This also includes promoting equality between women and men in decision-making.
Currently the organization has a female director as part of the board of directors of the company and has 40% as female employees.
The female employees are equally playing key roles in development, quality assurance, compliance, human resource, and finance activities.
Salcit is committed to creating a healthy working environment that enables its employees to work without fear of prejudice, gender bias and sexual harassment. Salcit clearly explains to all employees that sexual harassment will not be tolerated by any employee, manager, co-employee, customers, contractor, or anyone else who interacts with Salcit.
Our Business model primarily is B2B and B2G
Capex Model Description:
- Not applicable.
- No need for any special equipment or consumables. The application can be installed on a smartphone. If necessary the phone can be connected with an external microphone for ease of use. Overall including the smartphone and external microphone the total device cost will be around 100 USD.
Opex Model description:
- Only a fee for usage applies.
- Licensing (prepaid) model - bulk prepaid annual purchase of assessments.
- Per assessment (postpaid) model - pay as you go.
- Per user per month - unlimited usage - for longitudinal tracking of patients.
- Support staff, ASHA workers, ANMs. The only opex includes mobile sanitizers, use of masks (where appropriate)
- Organizations (B2B)
Combination of below
(1) Through grants
(2) Through revenues
(3) Raising capital
We have received grant money of about 300,000 USD in last 3 years.
And 10000 USD revenue in last year
Additionally, we have another 200,000 USD bootstrapped from founders
