Our real-life journey to launching a doctor-led start-up for critical-need revolutionary big data medical imaging technology
The journey to launching Cubismi started a few years ago when I, as a doctor, was trying to solve a critical-need for medical imaging to image cancer “heterogeneity.” Cancers are clonally diverse, meaning each tumor and metastatic focus can have very diverse biological subregions that behave differently with treatment. Subregions of tumors grow or respond to treatment, but we currently are not able to “see” the diversity of what’s happening inside the tissue. If we could, we’d be able to identify earlier cancer (and other diseases) and extend survival for patients using precision treatments.
I knew that the answer lay in a paradigm shift for medical imaging: harnessing the power of big data and machine learning. Cubismi submitted first-of-the-kind US and worldwide patents for medical imaging big data / machine learning systems (including neural networks) for “virtual biopsies” for which we have received broad claims from the USPTO.
Proof-of-concept for “virtual biopsies” methods using neural networks has since been established. These new in-vivo “biomarker” insights into tissues will become more powerful as big data systems grow. We foresee eventually answering questions such as: Is early cancer growing? Is a tumor a specific genetic subtype? Is a subregion of a tumor responding to treatment? Is a subregion of a tumor growing despite treatment?
But, how would we be able to organize all this new data? I also recognized that delivering these new insights would require an entirely new anatomically-organized correlative data structure — and the idea for the “My Virtual Body One File” was born (and, of course, patented). This medical imaging file of the future will take what are now widely dispersed and disorganized medical imaging data files and unify them, along with all the new insights, into a simplified single file. It will be the vehicle for advanced new 3D models of the human body, akin to a “Google maps” of the human body, that will allow new “street level” zoom capabilities of the new data-driven “virtual biopsy” insights.
Our team knew that these inventions could have a dramatic impact on human health, but getting intellectual property secured was just the beginning of the start.
Legacy technology pervades US healthcare (home of the fax machine), and medical imaging is still housed in technology and standards from the 1980’s. If you can remember the road atlases from the 1980’s, with each map separated and collated in large atlases, you can understand today’s legacy “picture archival and communication systems” (PACS) used in radiology. Each patient’s study contains many separate 2D images, with the number of 2D images per patient increasing rapidly. Today, one patient’s CT study can have over 1,000 images, and volumes are expected to reach 1TB of memory within the next few years. Further, new big data systems will require many other types of non-imaging clinical data that is currently siloed in many separate systems, including electronic health records (EHR).
Doctors currently act as human data integrators, spending valuable and expensive expert time in the mundane task of sorting images and surfing for data in the EHR looking for the most relevant information. Further, medical imaging often generates multiple layers of many different types of images, such as for MRI studies, thus creating many layers of redundant information on normal anatomy. In short, there is a lack of intuitive order for the data which forces doctors into menial repetitive work in sorting through less important information. Doctors are not able to simply focus of the pertinent image data needed for making clinical decisions.
In order to get our critical-need technology into healthcare, we thus realized that we needed to build a fundamentally new big data technology for medical imaging. We knew we could leverage the power of a start-up to gather a small group of top business, medical, and technology talent to help create a state-of-the-art big data technology — and “Cubismi” was born.
However, launching our big vision in the midst of our current healthcare crisis was not a simple journey. Of course.
The Wild Wild West
As a doctor-led start-up, we found ourselves facing opposing currents with the rise in new hype about “artificial intelligence” and big data. Industry groups touted patient data as the “new oil,” and we saw large (>$1 billion) financial mergers and deals around efforts to create or connect larger platforms that hold patient big data for what can be assumed to be for scaling, market dominance, and projected profits.1 2 3
By 2017, leading investment “experts” in Silicon Valley projected that radiologists and other types of doctors would be obsolete within five years or sooner by “artificial intelligence.”4 Private investment money poured into “artificial intelligence” start-ups attempting doctor-replacing AI, as well as new privately funded groups attempting to leverage AI automation for market “expansion” of clinical radiology services.5 6 These AI trends seem to have been in concert with the overall pressures on healthcare systems to capitalize on growing big with “value-based care” initiatives, which had already tilted the balance in favor of big: big healthcare centers, big tech deals, and big EHR vendors.
“Artificial intelligence” was touted as the answer to all of healthcare’s problems and new powerful industry-centric trade groups were born to explore use of automation in healthcare.7 Lobby groups representing big players poured money into efforts to influence Capitol Hill on how to direct use of powerful new fourth industrial revolution technologies in healthcare.8
But, something was lost. In this battle for big, the vital fiduciary relationship between doctors and patients was increasingly fractured.9 The choice of large healthcare systems to buy from big vendors further enslaved doctors to legacy EHR and PACS systems with poor UX designs — which has increased menial tasks and distracted doctors from focusing on their patients and clinical problem-solving. On doctor’s forums, being a doctor today has been described as being a “hamster on a wheel.” (with permission)
In our view, the diminishing of professional roles in medicine, in turn, has also caused other secondary effects on healthcare including: near exponential growth in the number of healthcare administrators and associated costs (#1 in the world, equal to 1.4% of total US economic output), high financial burdens on patients (medical debt is the #1 cause of bankrupcy in the United States), very low accuracy of data in EHR systems (eg. 89% of patients withhold medical data to protect their privacy), record-breaking rates of doctor burnout (>50% of doctors), and associated decreasing quality and safety outcomes (two-fold increases in adverse outcomes).10 11 12 13
The Tipping Point
Over the past few years, we have seen the impact of this new industry-centric personal data economy. The public has witnessed many negative ramifications of this burgeoning industry in the news, such as the new scandals with Cambridge Analytics and Facebook. We have seen the negative impact of industry (and specifically a monopoly of a few powerful companies) controlling and owning our collective personal data. Healthcare has not been immune.
The public has seen the high profile failures of big vendors working with large and complex healthcare systems, including some of the most famous US healthcare centers. In one case, over an estimated $100 million dollars was spent, with a net zero result implementing new computer intelligence systems to advance outcomes for cancer and other diseases.14 Despite grand plans, implementing complex new systems into healthcare across large systems by a large vendor proved far more difficult than anticipated. This failure seemed to have been in part caused by poor strategy by the technology company, but seems more so to have been caused by widely-publicized mismanagement and negligence by high-level healthcare system leaders.15 We have also seen complaints of ethics violations by large tech companies accused of blocking patient data housed within EHR systems, which elicited responses from government officials.16
In other healthcare system scandals, the impact of the industry-centric patient data economy led to the dismissal of notable physician leaders from their institutions. Patients reacted in news articles telling the healthcare system, “you betrayed my trust.”17 The enticement of massive potential financial gains got the better of these individuals, who breached what should be accepted standards (and regulations) for ethical use of patient data.
It’s Time to Re-Align
As a doctor in the middle of this imbalanced new big data world, I became very interested in new ethics considerations for applying machine learning in healthcare. I became involved with the IEEE Global Initiative for Ethics of Autonomous and Intelligent Systems and its collaboration with MyData.org. I am the chair of a new IEEE augmented intelligence standard for medical imaging that will incorporate new IEEE ethics standards, and am helping to lead an effort to start a new health data group under mydata.org. We realized that alignment with guiding ethical principles would be vital to creating the “right” big data technology for healthcare.
The IEEE’s Global Initiative for Ethics of Autonomous and Intelligent Systems is an effort to create new standards for ethical use of AI in society, having gathered world leaders on topics of AI ethics. MyData.org is a world-wide organization based on a whitepaper commissioned by the Finnish government which has spread across Europe, and is starting to expand across Asia and the US. In the MyData concept, individuals retain access and full control and agency over their digital data and personal identity in an individual-centric personal data economy.
It’s time to re-align back to the values of the medical profession by strengthening the core fiduciary relationships that protect patient interests (over industry interests), help foster high quality health big data, and assure the highest quality of care, including a robust doctor-patient relationship.
It’s Time for Great Design
According to mantra in Silicon Valley, all you need is an excellent team and front-end design doesn’t matter. We could not disagree more, especially for healthcare. Fixing healthcare is not as simple as designing the next app to plug into a flawed system. Creating new technology to help fix our broken system requires understanding core problems from a high level and from direct personal experience in order to find the best solutions, as well as finding a best means to incorporate these new well-designed innovations. It takes systems-level thinking and design combined with lean and iterative growth.
Design Thinking is a design methodology that’s increasingly being taught in healthcare today. It defines an iterative process for defining core problems in order to design best solutions. In this context, Cubismi is using Design Thinking to innovate best solutions for medical imaging as shown in Table 1:
We believe that the first step for use of machine learning in healthcare is consideration of ethics, and we have thus created “Guiding Principles.” For example, we will use machine learning (including neural networks) only for “augmented intelligence.” We will respect the end-user’s dignity and autonomy, and we will build a system that supports responsible behavior.
We will ensure a patient-centric patient data economy with new patient controls for the “One File” to allow patients proper privacy rights standards, including full control for use of their personal data, as well as improved access. We will replace legacy systems with powerful new technologies of the fourth industrial revolution. We will protect the doctor-patient relationship by centering our platform design around doctor-patient health data exchange. Trusted exchange of data with doctors, in turn, will lead to high quality data, which will power our big data system with accurate new “street view” insights in a new “Google maps” of the human body.
Last, we will flatten current barriers to implementing digital health innovation by using new digital health HUBs for our R&D efforts. These HUBs create a “blank slate” outside of current healthcare ecosystems to eliminate pervasive disconnects between healthcare buyers, doctor end-users, and patient end-users to provide an improved environment for successful development of new digital health innovation.
As our technology advances, we plan increasing validation of our core ethically-aligned design platform with adherence to new standards and policies, as well as incoming laws and regulations.
The Power of Health Big Data
As a doctor-led startup, can we get our critical-need live-saving big data technology into healthcare? Can we leverage the power of health big data for patients to drive a potential huge impact for human health? Can we use the power of health big data to keep you healthy?
Absolutely. We are at the tipping point.
US healthcare is in crisis with bad trends that continue to grow like a cancer.20 These trends have created an imbalanced empowerment of big players, those connected to big players, and those calling the shots for big players. Big industry and big healthcare systems, as well as closely-affiliated third-party start-ups and other groups supported by private equity, have worked to optimize health big data for proprietary use and big proprietary profits. And, sadly, it’s working. These companies and their affiliates are making billions off of our broken healthcare system. Despite enormous expenditures (20% of US gross domestic product (GDP)), quality outcomes for US healthcare rank very low compared to most developed countries.21 And, we are losing our greatest asset — our internationally-recognized top doctors and healthcare workers. The cost of this broken system to our society is enormous.
It’s time to re-align. It’s time for great design. We can put the power of health big data back into the hands of patients and doctors.
We can build new technologies and businesses that thrive while simultaneously assuring patients and society also thrive. We have an enormous opportunity as a start-up holding key intellectual property for creating the “right” big data technology, access to large financial investments, needed technology resources from partner large vendors, and new digital health HUBs available to us in today’s start-ups ecosystem to grow organically and make our vision into a reality.
As a doctor and as someone who has lost family members to cancer, I understand deeply how important this continued journey will be; this personal experience motivated me to start on this journey in the first place. We know that our new big data technology, in the hands of a driven team and the right doctor-led community, will make a dramatic impact on your life.
We know that the best things are born of the insights and motivations cultivated by the real-life personal journeys of people who can best design and innovate real solutions.
That’s the power of the start.