CRESCENDO accessible risk stratification

Enabling access to early awareness of pregnancy complications that are the leading causes of childbirth morbidity, and mortality with a highly-accessible, low-cost pre-diagnosis test. 

1) An onboarding questionnaire interaction with an instant-messaging chatbot.
2) Using optical sensors in phone cameras to detect blood pressure.
3) Retinal images to detect retinal changes for signs of retinopathy:
images to be taken regularly using the smartphone camera and analysed through ML image classification. Blood pressure will be measured through optical sensors already installed in patients’ smartphones and tracked daily. Continuous monitoring allows a baseline longitudinal assessment, and notifications can alert the patient to update retinal images and continuous BP tracking.

Delivery is solved through available and existing hardware: optical sensors in the phone camera along with instant messaging, or SMS. This not only cuts down the cost curve; distributing to more people increases the accessibility of prenatal need-for-care exponentially. 

Pre-eclampsia is the most common cause of both fetal and maternal morbidity worldwide. It can be a fatal complication to both mother and baby, causing strokes, liver rupture, as well as fetal prematurity and growth restriction. It is a multisystem disorder that complicates 2%–8% of all pregnancies, contributes to 15% of preterm deliveries, and 9-26% of maternal deaths worldwide.Globally, 76 000 women and 500 000 babies die each year from this disorder. Furthermore, women in low‐resource countries are at a higher risk of developing PE compared with those in high‐resource countries. Gestational diabetes is the most common metabolic disturbance in pregnancy and has increased by more than 30% over the last two decades, growing into a worldwide epidemic. Women with gestational diabetes are at higher risk of multiple foetal and maternal complications i.e. pre-eclampsia. Gestational hypertension is also a leading cause of maternal and fetal morbidity worldwide, contributing to 10–15% of all maternal deaths.This problem is exacerbated by insufficient antenatal provision in developing countries; in regions with the highest rates of maternal mortality, only around 50% of women received at least four antenatal visits, which are crucial to identifying complications in pregnancy.

It is a WhatsApp chatbot that is used for risk stratification of pregnant women at risk of pre eclampsia, gestational hypertension or gestational diabetes based on maternal risk factors, microvascular changes observed on retinal images, and tracking patients’ vitals i.e. blood pressure. 

By remotely monitoring pregnant women, it not only widens access to care in under-resourced communities, but also helps conserve resources by redirecting personnel to the identified high-risk populations and identifying gestational complications at an early stage. This allows us to adopt a preventative approach through careful monitoring and conservative/medical management of high risk patients before severe complications result. This is more cost-effective than treating emergency presentations of pre-eclampsia or managing chronic complications, and significantly reduces patient morbidity and mortality rates.

The chatbot regularly requests patients for retinal images that will be taken regularly using their smartphone camera and analysed through ML image classification, whereas blood pressure will be measured through optical sensors already installed in patients’ smartphones and tracked daily. In addition, an onboarding questionnaire will assess maternal risk factors for developing these complications  (as proposed by FIGO, NICE and ACOG). These three sources of patient data will be collectively analysed to identify high-risk groups of pregnant women.

Our solution serves to monitor women in both resource rich and resource poor communities throughout the course of their pregnancy, especially in the first two trimesters; as the complications mostly develop in latter halves of pregnancies. 

We will first be validating the test through commercial IRB trials, then licensing to women’s health clinics, IVF fertility clinics and ob/gyn departments in medical institutions after clinical approval. The first initiative will be piloted in 4 health centers and will be refined based upon user feedback. 

During the pilot period, we will be iterating the user interface and validating the accuracy of data collection and prediction. We will do this by engaging with medical professionals in the field of obstetrics and gynaecology, as well as with pregnant women.  Throughout the iterations, we will engage our initial users by asking for their feedback and learning from their data, thus creating a strong feedback loop that would better our solution to fit their needs. After the initial roll-out of the pilot, we will be refining the go-to-market implementation strategy based on user responses and the results of A/B testing across features of our data collection interface.

Pre-eclampsia, gestational diabetes and gestational hypertension are serious complications that afflict millions of pregnant women worldwide, in addition to engendering adverse outcomes in their offspring. These conditions disproportionately affect pregnant women in underprivileged communities, and incur hefty medical costs that stretch limited resources. By providing an easily-accessible, low-cost platform for risk stratification which can be accessed remotely, we identify high-risk populations in need of urgent preventative care, and flag them to local clinics for personalised management. This reduces the burden of prenatal care on under-resourced clinics, and mitigates the financial costs of treating late-stage complications through primary and secondary prevention.

Our solution enables access to early awareness of pregnancy complications that are the leading causes of childbirth morbidity, and mortality through a WhatsApp chatbot. WhatsApp chatbots have proven to have immense reach in low and middle income countries. For example, MomConnect in South Africa and the COVID-19 chatbot by WHO which has reached millions of people via WhatsApp. Furthermore, our solution is a test that incorporates a questionnaire, optical sensors, and retinal images, allowing us weigh each of these for reliability for every user, and ultimately creating a new testing tool for pre-eclampsia. Moreover, cutting-edge computer vision is used identify signs of retinopathy and propensity towards pre-eclampsia through retinal images. Using existing hardware in smartphones, not only cuts down the cost curve, but by distributing to more people, increases the accessibility of need-for-care exponentially.

Utilising cutting-edge ML computer vision to identify signs of retinopathy and propensity towards pre-eclampsia through retinal images. Simply using existing hardware, not only cuts down the cost curve, but by distributing to more people, increases the accessibility of need-for-care exponentially. Moreover, combining data from questionnaires (user-input) as well as from sensors on the phone, and images, allows us to tune the weights of each of these inputs for every user based on reliability of input, to successfully identify high risk patients.

Machine learning technology is widely used in healthcare solutions, and we could find supporting evidence from research papers that our approach is viable. 

In terms of validation of our predictive and diagnostic test effectiveness, we will be approaching 79 fertility clinics in California. Partnerships with fertility clinics in the area, where raising their successful pregnancy and delivery rates (SART) are mutually targeted goals, as our technology can lower the risks of pregnancy complications with early detection. 

The next targeted approach is to conduct pilot studies in OB/GYN departments of hospitals in Hong Kong, China, US, UK. As of Jun 2020, we currently have 5 pilot locations available to begin. To engender our focus on humanitarian impact, the crux of the solution is to deliver access for the highest-risk associated populations. According to research, this will be the targeted populations of pregnant women in South and Middle America with the highest incidences of pre-eclampsia and related complications, expanding accessibility to those who need it the most. During a 2019 survey fielded in nine Latin America countries, up to 97 percent of online shoppers in Argentina, purchase on smartphones. With online shopping made so accessible, pregnant women can make use of the existing channels, and have access to diagnostic checks through the phone, an interface they are familiar with having constant contact and feedback. 

Observed incidence of preeclampsia in the Swiss PRADO study compared with worldwide and regional incidences.

We will first be validating the test through commercial IRB trials, then licensing to women’s health clinics, IVF fertility clinics and ob/gyn departments in medical institutions after clinical approval. The first initiative will be piloted in 4 health centers in and will be refined based upon user feedback. 

During the pilot period, we will be iterating the user interface and validating the accuracy of data collection and prediction. We will do this by engaging with medical professionals in the field of obstetrics and gynaecology, as well as with pregnant women.  Throughout the iterations, we will engage our initial users by asking for their feedback and learning from their data, thus creating a strong feedback loop that would better our solution to fit their needs. After the initial roll-out of the pilot, we will be refining the go-to-market implementation strategy based on user responses and the results of A/B testing across features of our data collection interface.

In terms of regulatory approval, FDA licensing. We are currently engaging the advisory with an IP medical device lawyer. 

In terms of technological feasibility, our machine learning software requires the hiring of more software engineers and developers for our team recruitment and expansion. There is sufficient literature to support the application of the retinal imaging and existing hardware technology are simply smartphones and optical sensors. We rely on the scaling of software.

Full-time: 4 (Founding team)

Part-time: 3 (Undergraduate research interns)

We are a team of medical, business, and engineering women with support from Stanford University, Cambridge University, The University of Hong Kong, and The Hong Kong University of Science and Technology with 4 pilot locations and a team of industry advisors in the MedTech business. 

Potential collaboration with the MIT ecosystem includes:

The MIT Center for Gynepathology Research to build relationships with OB/GYN clinics and fertility IVF clinical partners. MIT’s Myers Lab and the work of Prof. Kristin Myers at Columbia and MIT. We are interested in the pre-diagnosis aspect of the Crescendo predictive risk stratification in cohesion with Prof. Myers’ work on childbirth mechanics of the cervix and uterus. 

Crescendo has a clear mission: 

1. Improved patient outcomes

2. Reduced barriers to healthcare accessibility

Key audiences:
In terms of impact: pregnant women with restricted access to constant health monitoring.
In terms of revenue, we will be licensing our test to women's health companies, IVF fertility clinics and ob/gyn departments in medical institutions. 

This is made possible by:

Step 1: Validation

Validation through IRB commercial testing of pregnant women to prove the safety and accuracy. Metrics to be defined during the study include the number of many patients we require, and level of confidence from our data collection. Further R&D to be planned for pilot programs at OB/GYN departments of hospitals. We are in the processing of engaging The University of Hong Kong, Queen Mary Hospital, Cambridge University School of Medicine, Stanford University Department of Ophthalmology. 

Step 2: Licensing for commercialization
Our key distribution channel are fertility clinics. 
Licensing to women’s health companies and IVF fertility clinics.

Our risk assessment technology goals of lowering fetal morbidity and mortality are in alignment with the goals of IVF Clinics, where they measure the SART success metric (sart.org). In partnership, increasing their SART score by providing the Crescendo early pre-diagnostic test to pre-empt complications for their user base. 

B-corp approach and philanthropy focus: e.g. for every $20 we charge for the test, we will contribute to expanding technological availability of internet networks and smartphones in rural communities to enable use of the Crescendo test services. 

Budgeting for Validation:

• Estimated cost of commercial IRB and timeframe, continuous study over the first trimester of pregnancy

  • E.g. $300 IRB cost per test

• Stages of approval for each test (to be refined at the next stage due to time constraints)

  • E.g. 8 months, 200 pregnant women, $800 remuneration for participation over continuous daily input of health data onto the Crescendo interface. 

We're an all-female team of multi-cultural, multi-disciplinary engineers, entrepreneurs, and medics who see an opportunity to work together, innovate, and make a difference for women and girls to deliver the largest positive impact. The MIT Solve platform gives us a starting boost for this project to take off. 

Potential collaboration with the MIT ecosystem includes:

The MIT Center for Gynepathology Research to build relationships with OB/GYN clinics and fertility IVF clinical partners. MIT’s Myers Lab and the work of Prof. Kristin Myers at Columbia and MIT. We are interested in the pre-diagnosis aspect of the Crescendo predictive risk stratification in cohesion with Prof. Myers’ work on childbirth mechanics of the cervix and uterus.