Customer Highlight: How Rune Labs is Improving Parkinson’s Patients’ Quality of Life Using Sensor Data Collected with InfluxDB
By Caitlin Croft / Dec 20, 2022 / InfluxDB, Community
Caitlin: Tell us a little about yourself and Rune Labs.
Carolyn: I studied neuroscience in college — while there I did computational research and learned to write code in an academic context. After college, I worked in the biology and neuroscience space, then joined a few startups as a software engineer. In 2019 I was looking for a new job and discovered Rune Labs. At the time, it was a small company and a perfect combination of two of my passions: neuroscience and software engineering.
Rune Labs is a software and data analytics company for precision neurology, supporting care delivery and therapy development. We are collecting and processing data from various sources including a mobile app, sensors implanted inside of patients, and from patients’ wearables. We are primarily working in the Parkinson’s medical research and treatment fields. We were lucky to gain these partnerships early on; there’s a lot of potential because of the deep brain stimulation (DBS) devices used to treat the symptoms and side effects of Parkinson’s. Our StrivePD platform is our care delivery ecosystem for Parkinson’s disease, enabling patients and clinicians to better manage Parkinson’s by providing access to curated dashboards summarizing a range of patient data sources, and by connecting patients to clinical trials.
Deep brain stimulation isn’t new technology, but there have been a lot of recent developments; the newer ones are able to stimulate the brain and record neural activity. You can think of them like electrodes that are implanted into the brain. With advances in recording capabilities, there’s a lot of academic research and clinical potential.
Caitlin: You’ve been at Rune Labs for over three years — how has your career developed since joining?
Carolyn: Over a year ago, I moved into an engineering management role; while I’m not in the technical weeds as much, engineering managers at Rune Labs are still pretty technical. I write less code these days, but I’m kept busy with strategic technical planning, engineering, and architecture design.
The team I currently manage is focused on our ETL data ingestion pipeline — from the point the data is uploaded to the way it’s processed, and the query API’s. A lot of the sensor data coming in is numerical time series data, and this is why we turned to InfluxDB. We call it numerical time series data because the data has a timestamp and at least one numerical value (there may be multiple values). Once the data is cleaned up and extracted from the raw data, it’s used in our platform and stored in InfluxDB. My team is in charge of the query APIs that expose the data to our users.
Caitlin: How is Rune Labs helping Parkinson’s patients?
Carolyn: Our platform collects real-time data that helps patients better understand the day-to-day lives of those with Parkinson’s. The data is ingested from DBS devices as well as our iOS app, and Apple Watch integration. Rune Labs partners with device manufacturers to develop software that helps them make better use of their data. We have a web portal for clinicians where they can view the data that their patients have asked us to share with them.
Rune Labs’ goals include:
- Improve patient treatment plans - present patient data that helps improve data-driven healthcare decisions.
- Better understand patients’ symptoms - Parkinson’s has a huge range of symptoms that are complicated to track.
- Become data-driven using real patient data - reduce the burden of tracking how patients are doing, and simplify tracking how environmental and lifestyle (i.e. sleep and diet) choices influence their state.
- Streamline medical care - Create useful data visualizations to make doctor appointments smoother and more effective.
The DBS devices require major brain surgery and not all Parkinson’s patients elect or qualify for the procedure. The devices are very effective at treating the symptoms of Parkinson’s: they don’t cure Parkinson’s, but they can greatly improve patients’ quality of life. Once implanted, it takes time to tinker with the settings to ensure the patient is getting the most benefit. It takes a team of specialists time to dial in the best settings for each patient; there’s a lot of back and forth between the patient and their clinical team.
Deep brain stimulators are electrodes implanted directly on patient’s brains. They’re somewhat similar to pacemakers. Pacemakers are closed-loop systems where most detect what’s happening and stimulate the heart based on the electrical activity. However, approved DBS devices are not closed loop systems; they are either on or off. You can adjust the settings, but they can’t (yet) detect what’s happening in the brain and stimulate the brain appropriately. Making them adaptive is the cutting edge of research right now.
Rune Labs helps facilitate the research side and make it easier to develop the algorithms needed to make DBS devices closed loop systems. The data is currently stored on the DBS devices until the patient is in their physician’s office; there, a special tablet is used to extract the metrics from the device. An advantage of the devices is that patients can share with their doctors not just their current state while in the clinical office, but also a longitudinal snapshot of the patient’s state as recorded by the patient.
Caitlin: Tell us about Rune Labs’ InfluxDB implementation.
Carolyn: We needed a tool that could collect time-stamped data and query it. We’ve been using InfluxDB OSS for a while and after considering other tools, we ultimately picked InfluxDB Cloud. We needed to upgrade to InfluxDB Cloud as we knew we were going to need to scale as we prepared to analyze more patient data. We are using InfluxDB to provide customers with visibility into sensor data collected from DBS and wearables.
We are using InfluxDB for all of our production data. We collect numerical time series data from a variety of devices, including electrical activity from deep brain stimulation devices. We take this raw data and clean it, normalize it, and store it in InfluxDB. We’re using the Go Client Library for read and write. After writing directly from a Go app, the data is used in the dashboards and visualizations that our users see. We want to have pre-built visualizations based on what we know is going to be most useful for clinicians. We don’t want to expose or force them to learn a query language. Eventually our data is sent to AWS for cloud storage.
Internally, we use InfluxDB’s UI and Notebooks to develop new queries. We’re looking at the raw values as that’s the most useful for development. Typically, we’re in the UI looking at output tables and ensuring they look the way they should.
Rune Labs takes security and privacy seriously. This is reflected in the way we store data in InfluxDB. Our data schema is quite simple: time series are tagged with one opaque ID, which is not directly linked to the patient’s identity. In a separate database in our platform, we store the metadata that links the patient to their time series data. This allows us to control the flow of information throughout our architecture: we are able to identify where in our platform there is sensitive information, so that we can be particularly cautious about the way that part of the architecture is accessed and adhere to privacy best practices.
Often with time series projects, data granularity becomes less important as time goes on. However, our internal team of neuroscientists wants to have a rich historical dataset, with as much granularity as possible. They often want all of the raw data metrics; they don’t want averages over time. As a result latency is an ongoing important project for us as we need to make sure clinicians are able to access the portal and graphs.
“We love that we can rely on InfluxDB Cloud because it’s horizontally scalable. It’s self-hosted and we don’t have to worry about our constantly growing data set.”
Carolyn Ranti - Engineering Manager, Rune Labs
Caitlin: What’s next for Rune Labs?
Carolyn: We’d like to help enable patients to be able to look at their data themselves, rather than needing to go into a doctor’s office to learn more. We know that not everyone is interested in looking at their state, but there are some who are really interested in the data collected by the devices and what’s happening with them.
Currently our platform is very tailored to Parkinson’s, but there’s a lot of opportunity to expand from a software engineering standpoint. Just this year we announced that we are partnering with Coastal Research Institute to help identify biomarkers (i.e. demyelination) of Multiple Sclerosis lesions in patients who have spinal cord stimulators. Demyelination is a standard symptom of MS when the patient’s immune system starts attacking nerve fibers which causes brain and spine lesions.
From an InfluxDB standpoint, we want to be able to make use of InfluxDB’s profiling capabilities as we want to ensure the queries we’re writing are performant.