MIT Solve

Solution Overview

Solution Name:

GGO AI

One-line solution summary:

Reducing the time needed to diagnose the COVID-19 in less than 2 minutes with decision support tool using Artificial Intelligence (AI) to de

Pitch your solution.

Analysis of a Pulmonary CT Scan result to identify Ground Glass Opacity (GGO) was carried out by a radiologist and takes up to 15 minutes. During the Covid-19 pandemic, Pulmonary CT Scan results from patients indicated Covid-19 increase but we have limited medical personnel who can analyze Pulmonary CT Scan results. What we offer to solve this problem a decision support tool using AI's ability to detect GGO in patients suspected of COVID-19, reducing the time needed to diagnose the COVID-19 in less than 2 minutes. This system cuts the time that use to analyze Pulmonary CT Scan results up to 86%. The analysis process, which usually takes 15 minutes to complete, will be shorter of only 2 minutes only. Thus, patients can get the medical treatment needed based on their condition as soon as possible.

Film your elevator pitch.

What specific problem are you solving?

Early diagnosis of COVID-19 is essential for the treatment and control of the disease. One of the methods used to diagnose this disease by identifying of Ground Glass Opacity (GGO) through Pulmonary CT Scan. GGO is the most common finding in COVID-19 infection. This method is more reliable, practical, and fast for diagnosing COVID-19. Compared to other means, the sensitivity of Pulmonary CT Scan is 98%, while RT-PCR shows a sensitivity level of only 71%. Analysis of a Pulmonary CT Scan result was carried out by a radiologist. This process usually takes up to 15 minutes. During the Covid-19 pandemic, so many Pulmonary CT Scan results from patients indicated Covid-19. Meanwhile, we have limited medical personnel who can analyze Pulmonary CT Scan results. In 2017, the ratio of radiologists per 1000 population in Indonesia only reached 0.01. Along with increasing cases of COVID-19 in all regions of Indonesia, this condition made our world at risk.

What is your solution?

GGO AI was designed to be use as a decision support tool using AI's ability to detect GGO in patients suspected of COVID-19, reducing the time needed to diagnose the COVID-19 in less than 2 minutes. Analysis of CT results using AI is one of the promising techniques that makes it easier for radiologists to diagnose COVID-19; meanwhile, many countries currently use artificial intelligence (AI) to assist health workers in handling COVID-19 cases. The system we will develop will help to shorten the diagnosis time to reduce the workload of doctors, maintain the accuracy of the analysis of CT scan results, and help to determine the priority scale of patients. This system cuts the time that use to analyze Pulmonary CT Scan results up to 86%. The analysis process, which usually takes 15 minutes to complete, will be shorter of only 2 minutes only. The confidence level of our software will measure the impact. We plan to achieve confidence level of over 90% when the project finished.

Who does your solution serve, and in what ways will the solution impact their lives?

In this current condition, number of COVID-19 suspected cases in Indonesia (ODP/people under monitoring) per date 16 June 2020: 41,369 person while the number of COVID-19 patients under surveillance cases in Indonesia (PDP) per date 16 June 2020: 13,574 person. This is the number of people who urgently need health care through a quick diagnosis of Covid-19. That is why our target customer is a hospital that has a CT-Scan service so we can help them by providing an AI based system to analyze CT-Scan results in a shorter time. Thus, patients can get the medical treatment needed based on their condition as soon as possible.

Explain how the problem, your solution, and your solution’s target population relate to the Challenge.

Health security during a pandemic is about ability to make an early diagnosis so an appropriate medical treatment can be done. Indonesia has a small ratio between radiologists and population, making it difficult to meet those needs. Moreover, during the pandemic, the number of people and patients under surveillance continues to grow and requires testing as early as possible to get a diagnosis. Estimated, there are at least around 53,000 people who need an analysis of the results of their CT-Scan. With our solution, faster diagnosis and medical treatment is possible, even though Indonesia has a limited number of radiologists.

In what city, town, or region is your solution team headquartered?

Denpasar, Denpasar City, Bali, Indonesia

What is your solution’s stage of development?

Concept: An idea being explored for its feasibility to build a product, service, or business model based on that idea

Who is the primary delegate for your solution?

Winston Wilson

More About Your Solution

Which of the following categories best describes your solution?

A new application of an existing technology

Describe what makes your solution innovative.

Many countries currently use artificial intelligence (AI) to assist health workers in handling COVID-19 cases. However, the use of AI to analyze CT scan results is still not widely used officially in health institutions. Though this has been proven scientifically in several academic journals.

the thing that distinguishes us from other companies that develop similar tools is that we develop this tool, not only as a diagnostic tool for covid but also other lung-related diseases. For this reason, we use various data sets from both COVID patients and other lung diseases. it aims to shape the sustainability of the system going forward, regardless of the pandemic situation.

Describe the core technology that powers your solution.

We use Artificial Intelligence as our core technology. We train deep learning system using collected datasets from CT-Scan results from hospitals, and online databases such as https://github.com/UCSD-AI4H/COVID-CT , and http://medicalsegmentation.com/covid19. Then, we process the data to separate the warp in the image from the CT Scan to extract the area of interest that differs from the positive and negative classes. This data is used for the AI training model and will be validated by a radiologist.

Provide evidence that this technology works.

We refer to several academic publications and studies regarding the use of AI as a decision-making tool in the world of Health. Specifically, regarding its ability to identify GGO through analysis of pulmonary CT-Scan results. In a research entitled Performance of Deep-learning-based Artificial Intelligence on Detection of Pulmonary Nodules in Chest CT (Li, et al, 2019) shows that the level of AI sensitivity in detecting GGO is 99.1%, far higher than a radiologist, which is 43%.

Please select the technologies currently used in your solution:

Artificial Intelligence / Machine Learning
Software and Mobile Applications

What is your theory of change?

To develop the GGO AI system, we first collect CT-Scan results data sets. After signing the agreement with the hospital for collecting CT-Scan samples datasets, we also conduct training for data collectors related to technical matters needed for the data collection phase. These data sets are needed to be able to do AI training and modeling until we have a system that reaches a confidence level of up to 90%.

Select the key characteristics of your target population.

Rural
Peri-Urban
Urban

Which of the UN Sustainable Development Goals does your solution address?

3. Good Health and Well-Being

In which countries do you currently operate?

Indonesia

In which countries will you be operating within the next year?

Indonesia

What are your goals within the next year and within the next five years?

We hope that in the next year, the system we have built has been validated and can be implemented in hospitals in Indonesia. in one year of its implementation, we target our system to be able to analyze up to 3100 scans per month per hospital with a confidence level up to 90>#/span###

What barriers currently exist for you to accomplish your goals in the next year and in the next five years?

Financial: Funding for system development

Technical: need more CT Scan results data sets for AI modelling

How do you plan to overcome these barriers?

Financial barriers: applying at another funding or grant program

technical: digging information about ct-scan data sets that available online while start to build the paretnership with hospital to collect datasets

About Your Team

Your Business Model & Funding

Partnership & Prize Funding Opportunities