6/5/24 - Episode 1 - MARTHA

Eric Snyder: Welcome to the first online seminar in our monthly series! Today we're going to present to you a live demonstration of Martha. I'm Eric Snyder, and I'm the executive director of the Technology and Innovation Group here at the University of Rochester Medical Center and the Wilmot Cancer Institute. Leading today's demo is our lead engineer and principal developer Scott Cunliffe. Scott was actually one of the very first developers we hired when our team was forming six years ago, and his expertise and code have been instrumental in bringing Martha live for these webinars. We are hoping to keep a lot of these between 12 and 15 minutes with some questions at the end to kind of keep this during a lunch break, so without further ado, I'll hand it over to Scott.

Scott Cunliffe: All right, thank you, Eric, for that introduction. Welcome, and thank you for joining us today as we discuss one of our newest apps: Martha. In the next 15 minutes or so, we will cover the following topics: what Martha is and our motivation for creating it, some of its key features, a live demonstration, current and future developments, a shout-out to the team, and finally, we'll provide contact information and have a question-and-answer session.

What is Martha?

Martha is a custom-coded analytic suite integrating tumor molecular testing from external and internal NGS Lab Partners. In order to do molecular testing, we send out our specimens to our next Generation sequency Lab Partners where they analyze biological markers in the genome and proteome. Martha is built on our Hyperion technology foundation, and Hyperion's the backbone of Martha as well as our other tools. Hyperion is an enterprise-level data integration platform consisting of a federated data warehouse with integrated data governance, ticketing, and analytics. Martha also uses complex multi-source data amalgamation routines to curate and deliver data to be used for oncology decision-making. Each of our NGS Partners have a different way of naming and structuring data; we transform this complex data into a simple-to-use tool. Martha has fast data integration of new Laboratory Partners via custom interface library, and we can typically have new Lab Partners added within two weeks. We have real-time data querying, which empowers patient care and research, including clinical trial planning and enrollment, so not only does Martha work in real-time, but having discrete data means we can also query specific data points not currently included in Martha. This provides an additional level of data analytics.

Why the name Martha?

Our team set out to try to come up with a name for this tool based on the specifics of what we wanted to develop. Our clinical director Dr Erica Ramsdale came up with the name Martha and its acronym. Martha Chase was involved with the Hershey Chase experiments, which determined that DNA transmitted genetic information. In 1969, Alfred Hershey won a Nobel Prize for this discovery, but Martha Chase was excluded. We thought this would be a good way to honor Martha, and typically, for our naming conventions, we try to find someone who's excelled in their field of work. Actually, Martha Chase herself briefly worked here at the University of Rochester.

What was the motivation that encouraged us to develop the Martha application in the clinical setting?

Previously, data was very difficult or impossible to find in the EHR (Electronic Health Record). Now, all the data is in one place for one patient or even your whole patient panel. Typically, to view a report in the EHR there's several steps that you have to take. For example, navigating the EHR to locate the results could be tricky and frustrating, but we make reviewing the results fast and easy. Also, oncologists were requesting the ability to search genes in alterations. This was the number one feature that doctors were asking for, mainly the tumor boards and disease working groups. Previously, it was a manual, slow process for identifying targeted treatments and clinical trial options. With this new app you get rapid identification of all eligible patients for research, replacing the previous process of manually scraping EHR and external lab PDFs to collect data. Now we've got the aggregate, real-time repository discrete data in a SQL database. Researchers can now prioritize their time on analyzing the data. For example, a doctor could look up patients that have the H1047R mutation in the PIK3CA gene with just a few clicks of the mouse. Martha helps doctors adapt their clinical trials, studies, and strategies immediately.

Key Features

Some of the key features of Martha are (1) an intuitive interface: you can search by gene, physician, patient, etc. (2) data is discreet and in real time: one click-access to the full report and (3) multiple labs are integrated into Martha: there is a plethora of information that is drawn in to create more accurate results. So, as you can imagine, bringing this large collection of disparate data from lab to app presented several challenges. Ultimately, we make this complex process look easy.

Live Demo

We’ll jump over to a live demo. As I mentioned, it has an easy-to-use interface: you have search capability where you can type in a keyword. By default, Martha's “Sort By” box is sorted by report date, but you do have the option to sort it by other fields, and by default the “Sort Order” box is sorted in descending order. For this demonstration I will search on our clinical director Dr Erica Ramsdale’s patients, and pretty much instantly the results are returned back. As we look at this we notice that the key word that we typed into the search box, in this case “Ramsdale”, is highlighted, and for every row that you roll the mouse over and call attention to, you’ll still be able to clearly see the original highlighted term. What we've included in the results set here is our patient name, the MRN (many of our sites don't use the Enterprise MRN; we'll look up and bring back all the EMRNs, so our users don’t have to worry about the other MRNs in the system), the ordering physician and the report date, and also the lab. The next four columns, with gold backgrounds, are the biomarker results- first one being tumor mutation burden status, the second being the score, the third being the micro satellite instability mix match repair status, and the fourth being circulating tumor DNA. Next are the genomics finding column. Any of the genes that have findings will be returned this particular field. We get the report ID and if you choose to look at the full report you can just click on the corresponding blue button on the far right- in this case the Foundation One report. If, for example, you want to look at a specific gene similar to the case of what I mentioned earlier, you can type “PIK3CA” in the search box and hit enter. Then you’ll get back the results displayed on the page: a whole cohort of patients that have that particular gene with a particular finding in it. That's Martha.

Current and future developments

We plan on adding labs and while we are currently in the process of doing that, we're also going to be adding gene alterations and the ability to search mutations including blood tumor data. We'll be integrating with our Gene Search app and other Hyperion applications, and we'll also incorporate clinical genomics and some of the areas will include population health, social determinants of health, social vulnerability index, and we’re going to be adding clinical decision support to help with therapy options, drug side effect research, and then survivability. Lastly, we'll add in some generative AI.

The Team

I’d like to acknowledge the team’s dedication and work towards the advancement of patient care. In the development of this Martha app, there’s Eric Snyder, who introduced me. He's our executive director providing strategy and tactical guidance. Dr Erica Ramsdale is our clinical director, who also provides strategy and clinical guidance. Scott Paoni is our head of product. JC is our research data engineer on this project; he is responsible for the design, the look, and the feel of the Martha app. Emily Strong is a research data engineer and provided the backend support bringing the data that we see in Martha to reality. You can request additional information on Martha or any of our other Hyperion based apps at our team website, https://www.urmc.rochester.edu/cancer-institute/research/informatics.aspx.

Eric: Thanks Scott! I also want to extend my thanks to everyone for joining us today; I just want to mention this webinar as well as the transcripts and everything in the future series will be available on the main site or you can search us on whatever platform you choose to search us on.

Q&A session:

Q: How easy is it to add in new NGS Labs?

 A: With the current set of libraries that we have for integrating data into Hyperion database, it will take about two weeks.

Q: Since this data is from multiple different labs, is the data standardized somehow?

A: Every lab has their own data schema and naming conventions, so when we get the data, we go through a routine that creates and matches up the general naming conventions that we came up with for Martha and populates all our tables with that general information. Once we're done with it every lab is pretty much standardized as far as how Martha displays it.

 Q: Do you have plans to open this up to other institutions?

A: Yes, we would love to share this with other institutions so by all means please contact us at our team website and we can talk about that

Thank you for joining us and hopefully we can do one next month as well!