IANN: Inference Analytics Neural Network-IANN

by Farrukh Khan

Inference Analytics Neural Network (IANN) improves physician productivity and quality by understanding notes, reports and other unstructured data in health systems through its proprietary platform.
Chicago, IL United States AI in Medicine BigData Informatics Equity Raise AMIA2019 challenge AMIA2019 Finalists challenge

All Team Company Patients Physicians Hospital Partners Mission Innovation Details Investor Info Due Diligence Docs Supporters Comments Updates

About our project

The problem we solve: Physicians do not have the time to write or read all the unstructured information about each patient. Physicians are not able to check/peer review their work. They spend extensive time documenting each case and they tend to burn out with the extensive amount of information they are required to absorb and read.

About our solution: Neural Network-based SAS platform improves physician productivity and quality. The platform is able to interpret information similar to what an experienced physician would be able to, generate and write impressions and conclusions similar to what an experienced physician would be able to and is able to assist and reduce the burden of reading and writing notes. Radiology reports, Pathologist reports, Discharge notes as well as general SOAP notes wil be able to get processed through the IANN platform.

Progress to date:

The IANN infrastructure has already been developed. We have 3 pilots already, University of Chicago Medicine, Tahoe Radiology, and Chicago Teleradiology. We are also in discussions and have generated interest from Mayo Clinic, Johns Hopkins, Emery University as well as University of Wisconsin.

We are co-presenting our results with the University of Chicago at the upcoming RSNA. In addition, Amazon Web Services (AWS) will be hosting us as part of their marketplace algorithms   

About Our Team

Creator: Farrukh Khan

Location: Illinois

Education: Kellogg School of Management (Northwestern Un

Bio: Farrukh holds 25 years of experience in product management, strategy and general management in analytic focused companies, and has held key early stage roles in companies that grew into Billion-dollar enterprises including IBM’s Netezza, Information Resources, Inc (IRI), Informatica, MicroStrategy and Strategy.com. Most recently Farrukh was GM and VP of IBM’s Netezza and Big Data Platforms (IBM acquired Netezza for $1.6Bn). Prior to this, Farrukh held senior leadership and product management roles at IRI and Informatica and was one of the first founding Product Managers hired into Informatica. In addition, Farrukh became a co-founder of Strategy.com while working for MicroStrategy. Farrukh started his career at McKinsey & Company evaluating technology for fortune 500 companies. Farrukh earned a MBA from the Kellogg School of Management of Northwestern University and a Bachelors of Science in Electrical Engineering from Columbia University.

Hospital Affiliation: Data sharing agreement with the University of Chicago Medicine

Title: Founder & CEO

Advanced Degree(s): MBA

About Team Members

Utku Pamuksuz
Co-Founder and Chief Data Scientist, Ph.D.
Biography: Dr. Utku Pamuksuz Ph.D. is a data scientist with expertise in artificial intelligence, business analytics, applied mathematics, and machine learning. Utku currently teaches at the University of Chicago as a part-time faculty in its’ Master of Science Analytics Program. Prior to that Utku has also worked as a Senior Data Scientist at State Farm® and Grainger®. Dr. Pamuksuz’s expertise is highly valued globally and has spoken at numerous academic and professional seminars in Europe, Asia, and the U.S.A. in topics ranging from computational social science to predictive analytics in the areas of Management, Finance, Strategy, and Quantitative Marketing. Utku earned a M.Sc. in Computer Science from Northwestern University and a Ph.D. in Information Systems and Data Analytics from University of Illinois Urbana Champaign.
Title: Co-Founder and Chief Data Scientist
Advanced Degree(s): Ph.D.
Twitter:
LinkedIn: https://www.linkedin.com/in/utku-pamuksuz/

About Our Company

Inference Analytics, Inc.

Location: 222 W Merchandise Mart Plaza
1225
Chicago, IL 60654

Founded: 2016

Website: http://www.inferenceanalytics.com

Blog: http://farrukhinferenceanalytics.tumblr.com/

Twitter:

Other link: https://www.linkedin.com/company/inference-analytics-inc./?viewAsMember=true

Other link: http://www.inferenceanalytics.com

Product Stage: Ready

Employees: 3-5

How We Help Patients

In radiology we expect that IANN will be able to reduce errors in radiology, hence patients will have a lower risk associated with physician errors as a result of using our platform. However, our platform is not geared for direct use by patients. 

How We Help Physicians

Physician productivity should increase by as much as 20% by using our platform, this has been attested through private practice use. Their error rate should also decrease by at least 50%. We have been able to see 20% physician productivity improvement and are willing to work with additional providers willing to use the system for further validation. 

Physician productivity should increase by as much as 20% by using our platform, this has been attested through private practice use. Their error rate should also decrease by at least 50%. We have been able to see 20% physician productivity improvement and are willing to work with additional providers willing to use the system for further validation. 

How We Help Hospitals

Hospitals have the potential to increase throughput from their radiologists and to reduce errors and risk by using our AI platform IANN.  By using our AI platform IANN, we predict that radiologists will be able to see more patients since the impression suggestion will shorten the time spent in writing each report. In turn, hospitals should see higher overall throughput. Additionally, the quality of reports should increase, resulting in a reduction on the number of errors in reports, and risks associated with those errors. 

How We Help Partners

Partners with established relationships with large hospital systems will have the opportunity to bring us in and share the revenue generated. We are developing these partnerships with software companies such as Nuance and hence would welcome the chance to work with additional partners who possess established relationships with large hospital systems.

Partners with established relationships with large hospital systems will have the opportunity to bring us in and share the revenue generated. We are developing these partnerships with software companies such as Nuance and hence would welcome the chance to work with additional partners who possess established relationships with large hospital systems.

Challenge Mission

Affiliation(s)

Strategic Advisors:

Dr. Paul Chang - Vice Chair of Radiology at the University of Chicago Medicine

Dr. Khan Siddiqui - Faculty at Johns Hopkins, Radiologist and founder of Higi

Dr. David Paushter - Former Chair of Radiology at the University of Chicago Medicine

Key Milestones Achieved and Planned

1. Our platform IANN (Inference Analytics Neural Network) has been validated by a major academic hospital for accuracy where it is being piloted and has now been validated by two additional private practices. 

2. Our work with the University of Chicago in using IANN to impact radiology workflows is being featured at RSNA this year. We will be co-presenting our work with the University of Chicago at RSNA this year. 

3. We plan to have our first paying customer using IANN by the end of 2019. 

Our Competitive Advantages

Our unique platform is based on an elaborate corpus of health care which is proprietary and patent-pending called the narrative cloud, along with individual algorithms trained using millions of records obtained through a specific unique partnership with one of the leading healthcare institutions sets us apart. Our platform is producing results that are more accurate than any other team that has approached this area of radiology workflows, i.e. impression predictions.  

Barriers to Entry

We have a unique patent-pending architecture that sets us apart, we have built a partnership with a leading academic hospital that enables us to improve and fine-tune algorithms and models in a real-world setting that is hard to match andour ability to bring together a unique team of AI scientists that possess both language processing/generation abilities and the skill sets needed to produce a true AI platform is probably our biggest advantage.

Funding, Partners and Alliances To Date

To date, we have raised $580,000 from Bold Brain Ventures, 10 Pearls, and multiple Angel investors between August 2018 and now. We have also received a $100,000 non-dliutive credit from Amazon Web Services (AWS).  We are currently extending our partnership with AWS and will be introducing a selection of new algorithms through the AWS marketplace. We have another $500K circled for our next round with our goal to raise at least $1.5M funding in the next round.  Note that we are open to taking as much as $4 million in our next seed round.   

Revenue

We are pre-revenue at this point and expect to have a first paying customer this year. We are aiming to reach an estimated $600K in revenue in 2020.  We expect to achieve 10X growth the following year, reaching $6.2M in 2021 and $22.5M in 2022. We forecast an estimated gross margin of 87% and net margin close to 40%. Initial sales and growth will come from targeted private practice groups valuing our established academic affiliations and our published papers that will be presented at RSNA which will further our credibility and reach. 

Innovation Details

Intellectual Property Summary

The architecture and technique used, as well the infrastructure used for IANN is patent pending. Our patent filings are in:

  • Multidimensional corpus where every healthcare term is mapped into a 300 dimensional space called a "Narrative Cloud".
  • Individual algorithms for specific improvements in radiology workflows. 

Clinical Information

We are co-presenting our work with the University of Chicago Medicine at the upcoming RSNA. We have been able to get validations from the University of Chicago on the accuracy of our algorithms, our accuracy has been higher than 95% in comparison to human-generated impressions in certain specific modalities. We expect to continue to progress in improving the algorithms while we are advancing new use cases and algorithms related to those use cases. 

Regulatory Status

FDA clearance is not required as we are working with human-generated data, i.e. not using AI to interpret images directly but looking at the text that is generated by physicians and providing the results as a reference back to the physicians to correct or improve their conclusions.

How we will use the funds raised

We have raised $580K, any additional capital raised would go into patent applications and GPU cluster costs 

Thank You

Healthcare has been inundated with the maintenance demands of EHR's to the extent that physicians are spending an extensive amount of time managing and entering needed information into these systems. This focus on data entry as part of the care delivery process has negated the experience for both patients as well as their physicians and provider teams. The impacts from this are far-reaching ranging from productivity and quality improvements all the way to physician burn out. Already there are numerous cases that I have personally observed of physicians who have moved away from practicing because of the way that care is being provided. IANN is meant to be a Physician’s assistant/helper to make sure Physicians get back to providing care and not become stuck in entering or deciphering data. We believe that care will be significantly improved for both the physician as well as the patient as the focus moves away from data entry and towards care delivery.

Investor Info

Market Size

We estimate our Total Market Opportunity to be $1.7B built from four distinct areas where extensive documentation or narratives/notes are being generated:

  • Radiology 34,000 physicians generate 28,000 reports/year on average = approx. 1 Billion radiology reports/year.  We would charge $.50 cents per report = $476M  
  • Pathology 18,000 physicians generate 12,000 report/year on average= approx. 216 Million pathology reports/year. We would charge $2 per report = $432M
  • Discharge Notes 35 million hospitalizations/year on average generating a discharge note.   We would charge $12 per note = $420M
  • SOAP Notes for applicable specialties.  Approximately 4 billion notes across specialties (excluding procedural/surgical specialties)/ year on average.   We would charge $.09 per report = $360M

 

Projected 3 Year Growth

We expect to be servicing approximately 6,600 physicians, which accounts for approximately 19.4% of the total radiology market within 3 years. Generating revenues of about $22.5M, with gross margins of about 83%. 

Revenue Model

Radiology is our first focal market. Per our opportunity analysis, it is obvious that this is the largest segment that we are pursuing.

In radiology we are targeting the following customer segments:

  • Academic Hospital Systems - given our access to key leading advisors in Academic environments, we have received strong validation and acceptance for out work - Events such as the SIIM (Society of Information Informatics in Medicine) and RSNA participation as a presenter will allow us to get access to leading academic institutions. We expect to charge these institutions our standard fee of 50 cents/report. 
  • Strategically Targeted Small, Medium and Large Sized private practices to start off with. We have targeted key influencers in private practices for our pilot programs.  These key influencers have strong credibility and acceptance in the market.  By turning our existing private pilots into customers, we expect to be able to have strong references that will facilitate our acceptance by other prospects. We expect to maintain the .50 cents/report price point
  • Large hospital systems-- our goal is to use partners to penetrate the large hospital system world. Through reseller models that we are currently exploring, we will leverage incumbent vendors to penetrate these systems, leveraging their relationships in the process. The price point would remain at 50 cents / report. 

 

Competitors

  • Radi AI - this is the closest radiology competitor. Their platform is not able to produce the level of speed with which our impression calculations occur. Our advantage over them is our platform and overall technology architecture. 
  • Agamon health  - this is a general text-oriented deep learning platform. They are not radiology focused, hence not a strong direct competitor
  • Amazon Comprehend - We are partnering with AWS, and leverage their GPU's, additionally we expect to launch our algorithms in their market place. Their inability to get access to training data sets as well as a more general-purpose capability makes them more of a partner than a competitor. 
  • Optum - They are based on rules-based NLP as opposed to automated, deep learning-based models. There is a limitation on what a rules-based system can achieve.

We offer advantages in prediction, speed and quality versus our competition due to our comprehensive architecture which combines a unique healthcare corpus that we have trained, in addition to our individual unique algorithms that work for each specific use case.

Traction

Today we are being piloted at one of the leading academic hospital systems with a focus on radiology. We are presenting our work in impression prediction jointly with the University of Chicago at the upcoming RSNA. We have 2 private practices who have been using IANN, they have seen 20% productivity improvement by using the platform. We are expecting to convert these to revenue-generating customers. We have partnered with Nuance, the leading software platform for radiologists in report writing. We have been able to integrate with their API, which they do not expose to many partners. We expect to be able to leverage this partnership for further traction.

 

Due Diligence Docs

Please note that access to the company's confidential materials is limited. Click this button to request access from the Company and its representatives.


Supporters

Comments

Login to post your comment!
Click here to Login

Updates

    No updates found .

$ 1,001

pledged of $ 1,500,000 goal

  • 5 Days left

105

Interest
Score

0

Adoption
Score

Farrukh Khan
Founder & CEO
Kellogg School of Management (Northwestern Un

Proflle
Contact Me


Rewards All contributions are tax-deductible.