Dreams do come true! Not always when we want them to, not always in the way we would like them to, but they do, if we are consistent in our efforts and focus on realizing them every moment.
As for Cupid Chan, his dreams lay in the entertainment industry. He even won a singing contest as an undergrad in Vancouver, Canada, and an agent from Hong Kong invited him to join a media group afterward. However, after a 4-hour marathon debate with his parents, who believe the entertainment industry is risky and worthless and that he should focus on what he studied, and treat that as his career, he gave up his dream.
Temporarily.
After graduating, Cupid took his first software engineering job in the Greater Washington DC area. To stay competitive, he decided to go for further studies (Master’s) at Johns Hopkins. But his childhood dream still lingered in his mind. Soon he learned he doesn’t have to be in the entertainment industry to be an entertainer. He can combine what he learned from school with what he is passionate about and make it his career – hence he identified himself as Technology Entertainer – he develops technology that is interesting not only for geeks but also intriguing and beneficial to everyone. “I use this principle at my work as an employee, in my keynote as a speaker, in my lecture as a professor (I am appointed as a Senior Fellow at both University of Maryland and Johns Hopkins University), and at my company as an entrepreneur,” shares Cupid.
Since then, he has played different roles at different companies and become a consultant with some Fortune 500 companies as his clients. Because of his data and analytics background, he also played the Head of Big Data and Analytics role in one of those Fortune 500 companies – CSC (now it’s being split, and merged, and acquired with HP, and SRA, and GDIT) before he joined a company as their CTO. In his most recent venture, he took a leap of faith to start his journey and found Pistevo Decision to use technology to tackle the healthcare industry.
From Living the Problem to Creating the Solution
Cupid met one of the co-founders, Dr. Ferdinand Hui, at an AI event, where they spoke – he from a technology perspective, and Dr. Hui from a medical perspective. While Dr. Hui was concerned about helping more patients and that the current system was more pro-doctor, Cupid was eager to get a great use case to build on the technology that he knew. Dr. Hui brought on board his colleagues at Johns Hopkins, Dr. Harry Quon and Dr. Phil Phan, and Cupid brought his technology buddy Calvin Chiew to the pool and found Pistevo to build a patient-centric system. But the next question was: “How?”
They tried a couple of things but didn’t feel like that’s what a patient wanted. In the meantime, Cupid was diagnosed with hyperthyroidism and saw himself filling out redundant forms at different places (blood works, X-rays… etc.) with 90% of the content he had already filled in at other locations. It got him thinking, “For a patient-centric system, should I, as a patient, be the owner of my information, and if I want, I can press a button to share that with my doctors? Suppose a hyperthyroidism patient needs to go through that; what about patients with more complicated diseases like cancer?” So, they decided to build a system to help patients collect and share their data if required.
Next Step: A Solution that Safeguards Data
Sharing data sounds convenient. However, “data is the new oil,” and there are tons of value locked inside the data if we know how to harvest it. That’s what big companies do onwith customer data, while we, the data creator or owner, get no financial reward. Moreover, sharing data seems to invade privacy, a big no-no in this era. What can we do to let people harvest the value from the data while holding up data privacy, and even better, the data owner can earn money in the process? Because of his role as the Chair of the BI & AI Committee at Linux Foundation LF AI & Data since 2017, Cupid got chances to talk to industry leaders and learn that Federated Learning may be the solution.
“Even though everyone is talking about AI, traditional AI has at least two issues. (1) It collects data in a centralized location for a machine learning algorithm to run and build the model. But when the data is collected, data privacy is also already breached. (2) “Data is Fire.” Like sharing a fire from my torch with yours, I lose control of how you can use my fire right after that. You can keep passing along without my knowledge, and I cannot even “get it back.” Therefore, unlike tangible objects, the value of the data immediately drops the moment it’s shared because I cannot recoup it,” explains Cupid.
By using Federated Learning, one can turn the whole scenario around. Instead of collecting data in one place, the model is pushed out to where the data is located for training the model. Once done, the little improved model is sent back, without the data, to form the aggregated result. Hence, the data consumers achieve their goal because they have millions of quality data to train their models. Data owners still possess the data without losing privacy. And because data never leaks, data owners can now set a price for people to use their data.
“Wait, Federated Learning in Healthcare? What about calling it Federated Healthcare?” One Sunday morning, Cupid woke up with this idea. Then he searched online; there was no such thing. So, he registered the domain FederatedHealthcare.com to promote this concept. This will be the Marketplace for the data owner, i.e., the patient, who can share their data as in the old traditional way if they don’t care about privacy. But they will also provide a way for people to share data privately. This will be a place for those who need high-fidelity healthcare data for research or to train their machine learning model. And the result of the model can use the very same platform to identify potential buyers who can benefit from it. This ecosystem will allow them to unlock the value of healthcare data without sacrificing data privacy or losing data value to propel research and improve healthcare as a whole.
The Process of Innovation
Innovation is optimally attained when you know well enough to develop a solid solution but do not know too much so that predefined boundaries do not limit creativity. This is why Pistevo has founders and partners from different areas. Each contributes their expertise from their domain. Once the pain points are shared, others can ask “dumb questions” to inspire a better solution that may not even be considered if you are in the domain for so long.
“Success is built on top of many dumb questions because each dumb question leads to an incremental improvement heading toward the finish line. Moreover, the answers to those dumb questions are usually simple yet foundational. Without knowing them well is the same as building up a house on a sinking sand foundation,” shares Cupid.
Keeping up with Technology
As technology is dynamic, so should be your response to it, believes Cupid. “I still remember that the waterfall model was the gold software development standard while studying my undergrad. But now, everyone is adopting Agile. We are so spoiled with a quick turnaround. The average time a person will wait for a loading page is 4 seconds. If it’s longer than that, good luck to your site, as the user must most likely switch to something else. Nowadays, people do not have the patience to wait months or years to see the result. So, to make the solution scalable, the first thing is to make your solution “available” whenever the users need it,” he shares.
Unfortunately, users’ needs change over time quickly due to quick environment evolution. Hence keeping delivering value incrementally, hearing the feedback from the users, and incorporating enhancements into the next release is critical to making a solution scalable. The transition involves education and mindset changes. “But “feed your own dog food” by applying this theory to how to help our clients by showing them improvements time after time. Get them involved as part of the team in the process actively. This makes them co-own the product instead of just a bystander watching the game,” he says.
Challenging Time
The toughest time for Cupid was when he first switched from a technical to a managerial role. His inside pride said, “Now, you have a different status. Just hand off work to your team and enjoy being a manager!” Big mistake, in 2 senses. First, attitude: a manager is not the same as a leader. A successful business needs leaders who walk the talk, not a manager who talks the talk. Other people can quickly feel that, and you will soon pay the price as your team runs away because someone just shoveled work down their throats.
Secondly, quality and vision: Everyone has 24 hours every day. As an entrepreneur, you can and should spend more time developing the business, which means you will have less time to involve in technical build-out. How do you ensure the product quality is maintained with the vision you set? “As your team is an “employee,” it’s naïve to think an employee will have the same level of dedication as you, an entrepreneur. It’s not saying that they intentionally do a terrible job. But realistically, the commitment will never be at the same level as yours. Moreover, handing off may lead to a different direction that you, as the entrepreneur, envision,” he explains.
Therefore, before handing it off to your team, you should develop a repeatable process. A process that will guard against the quality of the work automatically, e.g., Continuous Integration and Continuous Delivery (CI/CD) in software development, and have a periodic meetup to ensure the outcome is appropriately aligned with the vision. Once you confirm that the team grasps your idea and builds trust gradually, you can then get let go more and more over time.
“There is no shortcut to building this foundational process. If you don’t do it sooner, you will do it later at a higher price. Handing off too soon without a proper strategy is wasteful of resources. It took a long time to reboot, which was how I overcame this costly mistake,” he shares.
Understanding the AI Wave
Explaining the latest AI rage, Cupid shares that just like Google has provided great search results since 20+ years ago, one still needs to sift through what is appropriate. ChatGPT returns a great language model’s response, but you are the one to judge whether it is accurate. Technology is to assist humans and replace those who totally rely on the result without the knowledge to see what’s right or wrong.
“I keep hearing this recently: ChatGPT is going to replace many humans at work, and many people will lose their jobs! My response: There was a similar concern when machines were first introduced in the industrial revolution, but the number of jobs per capita has increased since then,” says Cupid.
“New technology will make some legacy jobs and people who don’t want to change irrelevant. But it will also push society forward, and history has proven that time after time. Like it or not, this AI wave is coming. The question is not “Are you ready” but “How can you get yourself ready and stay relevant”?”
The immediate challenge, for the industry, as per Cupid,As per Cupid, the immediate challenge for the industry is how quick and agile you are to adopt the new technologies. Owing to AI development, changes are exponentially faster than we had ten years ago. While inventing those emerging technologies is significant, businesses should think about leveraging those technologies quickly to complement what they currently have.
The Path Ahead
Cupid and his team aim to build out the Federated Healthcare platform in the coming years. But they understand nothing can be achieved if they try to boil the ocean. Healthcare is a gigantic ocean with many unknown monsters. So, they plan to achieve this goal step by step. “Do one thing well, secure the ground, and expand from there. Therefore, instead of targeting ALL healthcare data, we must first focus on one,” he says.
Not so long after they started debating what to focus on, Cupid got a message from his dad that his mom got a stroke and was admitted to the hospital. That was when they still were in lockdown mode due to the pandemic. Traveling to Asia required quarantine for two weeks each way. But he knew the hospital took the MRI for his mom, and his dad has the CD. Having a co-founder as a stroke specialist, he immediately asked his dad to upload those results so that he could pass them along to Ferdinand for some advice.
But too bad it took them a long time, yet they still needed to find a way to share this information. Ferdinand further confirmed that even though a hospital may have a medical image system, no such patient-centric approach allows sharing as the laypeople expected. This was it! Medical Imaging was the first healthcare data they incorporated into their platform, and they released it earlier in 2023 for patients and doctors to try out. “We brand this platform Carealth – to streamline Care and Health for real,” shares Cupid.
“In the meantime, applying the same idea, we won our first NIH SBIR award for incorporating Social Determinants of Health (SDH) data for cancer patients. We finished the Phase I development in 2022 and expect the Phase II award to start in 2023. Based on recent research, addressing this issue helps reduce readmission rates for the healthcare system and save overall healthcare costs. Therefore, in the next few years, we will leverage the Carealth platform to expand into SDH,” he says.
Message to the Budding Entrepreneurs:
Cupid concludes our interaction with the following message for budding entrepreneurs:
“Analysis paralysis is the number one differentiator between a visionary and an entrepreneur. In the NIH SBIR award I mentioned above, we are also accepted to an optional I-Corps program. A critical lesson from this program is reaching out to your users and understanding their needs. Then incorporate the feedback into the product or service, and don’t be afraid to fail. Keep repeating this process to search for the right product-market fit. Just sitting and thinking your solution is the best one leads you nowhere. Taking action, listening, and improving is the positive cycle an entrepreneur needs to follow. By practicing this, we won the “Best in Show” Award in the I-Corps program!”
Quote: “Great Idea – Action = Failure; Listen humbly + Improve continuously = Success eventually”
Built on Trust
“Pistevo” means to trust or believe in. I realize that sometimes, it’s not because the technology is excellent to make a company successful. But whether people “trust” you. Building trust is hard, and breaking it takes a minute or less. One crucial way to build trust is to maintain transparency. I prefer open conversation, especially in the case of conflict, as I found out that conflict usually is because of miscommunication, which can be resolved relatively easily if we keep the channel open and transparent.
Dreams do come true! If Cupid can focus on realizing them, one day, a Technology Entertainer will become a Technology Celebrity!