🔬Can smart devices help prevent serious diseases long before symptoms develop?
🔬Can smart devices help prevent serious diseases long before symptoms develop?
Hello,
How do you come up with blockbuster ideas in your field? Don’t over-optimize on your ideas, but run many experiments. “When it comes to idea generation, quantity is the most predictable path to quality,” said Adam Grant in his bestselling book, Originals.
He shares examples of master creators such as Pablo Picasso. Though Picasso is renowned for a handful of artworks, he has produced more than 12,000 drawings, 1,200 sculptures, and 1,800 paintings.
I’ve started applying this learning by experimenting and producing more. I test out multiple concepts in my area of expertise through LinkedIn posts, newsletters, and networking conversations. I iterate on those that seem to resonate the most.
This newsletter will take you about 5 minutes to read.
I. Spotlight: 🔬Can smart devices help prevent serious diseases long before symptoms develop?
“Hey Alexa, how healthy am I today?”
Very soon, your digital assistant can answer this question just based on your voice. Actually, this technology already exists.
Today, smartphone apps can detect depression just by listening to someone speak a few sentences. Surprisingly, the language or the words spoken aren’t as important as how you say it.
Digital biomarkers can diagnose complex health conditions just from the human voice. AI can spot patterns in the tiny variations in voice by analyzing the pitch, energy, tonal quality, and rhythm. This is called the voice biomarker.
“We have many such digital biomarkers, such as keystroke dynamics, eye tracking, facial recognition, smell, and gait,” says Richie Bavasso, Co-founder and CEO of nQ Medical.
(Photo by Daniel Korpai on Unsplash)
The coming digital biomarker revolution
Today, we have digital devices all around us - smartphones, watches, and many other wearables. The sensors bundled in these devices collect data passively throughout the day.
When you apply machine learning algorithms to these massive data streams, detecting health anomalies can become a child’s play.
When you map health readings from smart devices to disease outcomes accurately and reproducibly, they can be considered digital biomarkers.
Digital biomarkers are enabling decentralized clinical trials (DCT) - trials that can be run remotely from the comfort of patient homes. “Recent research shows that a typical DCT deployment could shave one to three months off timelines in each phase of drug discovery,” says Andy Coravos, CEO of HumanFirst.
Here are three steps for researchers to adopt digital biomarkers in their clinical trials:
1. Define what to measure
“Start by defining the attribute you want to measure,” advises Coravos. “Let’s say you want to measure sleep. Your Apple Watch or Fitbit may not know if you really slept. Is sleep the number of minutes you were in bed or the bouts of deep sleep you went into? There are probably 35 different ways of measuring sleep.”
“Our team has compiled thousands of peer-reviewed papers, clinical trials, and regulatory filings,” adds Coravos. This can help researchers compare the different outcomes and pick the most appropriate one for their study.
2. Choose the device
While hundreds of devices are available in the market for health indicators, they may not be suitable for your study. Each device comes with technical guidelines on what it measures, how often, and how best to use it. These attributes must be analyzed and compared to pick the most appropriate device.
Atlas is a platform built by HumanFirst that helps researchers evaluate more than 1200 devices addressing 8000+ physiological and behavioral digital measures. OpenAtlas is an open-access version of this platform.”
3. Analyze and interpret results
The next step is to convert the raw data readings into meaningful patient outcomes. Take the example of nQ Medical’s solution that detects Parkinson’s disease by analyzing patient keystrokes on their smartphones and/or computers.
Bavasso’s team has trained machine learning algorithms to spot the level of motor skills degradation based on the user’s typing behavior. The algorithms generate a numeric score (nQi) between 0 and 1 to indicate the disease severity. This score is computed continuously every 90 seconds as users interact with their devices.
(nQi score (0 to 1) indicates the severity of fine-motor impairment measured from natural typing - Photo by nQ Medical)
Clinicians incorporate the nQi score into their current standard practice as a complementary measure that is more objective, granular, and accurate.
Thanks to digital devices, passive data collection, and intelligent algorithms, digital biomarkers can soon prevent diseases even before patients develop any symptoms.
II. Industry Roundup:
1. Webinar: Designing your winning data strategy
45 minutes | The Data Chief | Cindi Howson
Where should you start to design a data strategy? This chat with industry leaders presents critical elements of data strategies, how to socialize it within the broader organization, and the challenges in executing the strategy. Through one-on-one and panel discussions, Cindi helps answer many common questions on this topic.
2. Article: NYC mandates hiring algorithms to be audited for bias
12 minutes | Wired | Khari Johnson
New York’s City Council adopted a law requiring audits of algorithms used by employers in hiring or promotion. The law is the first of its kind in the country and signals how AI could be regulated. There is consensus building up on mandating audits or impact assessments for AI use. However, there is a lot of debate on how to make them effective and avoid “ethics washing.”
III. From my Desk:
1. Interview - Tom Redman: Dirty data = Messy models
13 minutes | Gramener
Curiously, everyone wants to do the model work but not the data work! Tom Redman and I discuss what drives the misconceptions about data quality and how it hurts organizations in this chat. Are new buzzwords like data fabric, data mesh, and modern data stack helping improve the data quality?
2. Article quote: 3 ways to improve omnichannel engagement in life sciences
9 minutes | MedAdNews | Bruce Carlson, Aktana
Companies cannot be successful by optimizing for a single channel. The interplay of communication channels - also called an omnichannel strategy is all the rage today. How can data analytics help lay a strong foundation here? I shared my views in this MedAdNews article from Aktana.
Have you tried the viral Wordle game yet? I got hooked. Thankfully, it caps at one word puzzle a day!
Thank you for subscribing and reading the newsletter. I appreciate your attention,
Ganes.
PS: Did someone forward this to you? You can subscribe here.
My Website | LinkedIn | Twitter | YouTube
I’m Ganes Kesari. I publish ‘Data-Driven Future’ to help understand how data shapes our world, explore key trends, and explain what they mean for you today. I speak and write to demystify data science for decision-makers and organizations.
Recent Issues: