Posted by Summary: This post covers (Reading time: ~10 minutes)
- HealthTech pioneers need to be on top of the rapid progress
- Even InsurTech pioneers (developers) need to be vigilant
- Future AI/ML developments in Digital Health
- AI Solutions for self-management of chronic diseases
- AI Solutions for Clinical Care – Prediction, Early Detection, Screening, and Risk Assessment
- AI in Surgery
- Future Healthcare Imaging infra development
- Clinical Imaging Systems & Research Imaging
- Imaging Molecular and Cellular Biomarkers
Digital innovation has bought many novel capabilities to the Healthcare sector. The last decade has seen exponential capability additions in Healthcare infrastructure. More significantly, medical imaging innovations are beginning to gain abilities that were not even conceivable not far ago. My previous learning blog covers the fast-paced innovations in the last decade. This part covers how future capabilities are evolving. My learnings rely heavily on the following two academic papers:
- Special feature paper, Advancing biomedical imaging by Ralph Weissleder and Matthias Nahrendorf, Center for Systems Biology, Massachusetts General Hospital and Harvard Medical School, Boston.
- Artificial Intelligence in Health Care: The Hope, the Hype, the Promise, the Peril; Michael Matheny, Sonoo Thadaney Israni, Mahnoor Ahmed, and Danielle Whicher, Editors
The sheer quantum and diversity of digital imaging capabilities innovations are overwhelming. But the conjunction of developments happening on multiple fronts is what is disruptive. Digital disruption is pushing boundaries in almost every sector. Together, these innovations have the potential to bring in massive step change. As a result, the future looks quite overwhelmingly exhilarating.
HealthTech pioneers need to be on top of the rapid progress
Digital pioneers developing Healthcare solutions have their challenges cut out for them. The simultaneous speed of transformation in multiple sectors means one needs to have a holistic awareness – every tool or tech deployment needs to accommodate the ever-faster changes.
What happened to the fledgling paging industry, being eaten for a snack by the android ecosystem, will happen more frequently now. Within the time it takes to evaluate innovation and complete development efforts to make it commercially viable, it can be obliterated by something 10X better.
A compounding factor is the legacy mindset of Healthcare Provider Mindspace (or the Healthcare professionals) and Regulators, for whom the fast-paced capability may be scary and even feel alien.
Even InsurTech pioneers (developers) need to be vigilant:
Insurance plays a significant role as a payer-enabler in the Healthcare system. As Digital Health is moving from indifferent (to a patient’s personal situations) illness protocols to proactive (customer-centric, not patient-centric after an illness becomes concerning) wellness protocols, the Insurance space needs to adapt too.
Insurers facilitate a risk-sharing platform to pay a manageable cost (premium) for possible future losses. In the case of Life & Health Insurance, these risks are demographic mishaps, accidents, critical illnesses, or even unfortunate untimely death. These are the very events that Digital Health targets to be customer-centric. Hence any consumer-facing InsurTech solution needs to plug in seamlessly with underlying innovations in the Digital Health.
InsurTech will critically depend upon the data feed from the HealthTech, enabling an economic structure and smooth payments, but more importantly, providing a symbiotic feedback loop to HealthTech. The fast-changing Healthcare capabilities will pose a similar threat of making a solution obsolete, given the time it takes to develop and adapt in a highly regulated Insurance ecosystem. A healthy person using InsurTech in proactive prevention protocols will be sure more challenging than a sick patient with less choice in the current system.
There is a fundamental challenge for the actuaries supporting life and health insurance. The Healthcare system transforming into a wellness partnership will lead to proactive interventions affecting the occurrence of the incidence. The diagnosis and treatment innovations further will more likely like 10X better at 1/10X cost, radically changing the survival impact and post-survival prognosis. The challenge for us is to predict the changes in the past trends and adjust the insurance costs.
More profoundly, the compounding impact of Digital disruption has the potential to create a massive wave of wealth; some digital enterprises now have financial clout bettering the national GDP of many countries. Every time a wealth wave has happened historically, it only has improved the healthcare support for the masses.
In a way, insurance fills the gaps in social contracts for the distribution of wealth in society. As these gaps transform/change contours with a new wave of wealth, so will the insurance needs. The insurance industry can adapt to changes that occur at a reasonable pace, but the traditional working models for long-term insurance fail at the current past-paced disruptions.
Now to the crystal-ball gaze:
This learning blog focuses on the areas covered by the IISc – Talent Sprint course, future exciting possibilities driven by AI/ML capabilities, and the innovations in Healthcare Imaging/technology. Again, these are symbiotic, the innovations feeding better and more data to AI/ML algorithms, creating compounding future possibilities.
Future AI/ML developments in Digital Health:
The Healthcare industry has been investing for years in AI/ML solutions. As a result, some promising examples of AI solutions have emerged. Still, scaling up in the highly regulated industry has been challenging, with fragmented data and privacy concerns and professional medical expertise unconnected with AI/ML frameworks.
It is also difficult to assess the impact of combined solutions. For example, an AI solution may become exponentially more potent if coupled with augmented reality, virtual reality, faster computing systems, robotics, or the Internet of Things (IoT). Some areas where AI/ML are making transformations are summarised below.
AI Solutions for self-management of chronic diseases:
Self-management can be assisted by AI solutions, including conversational agents, health monitoring and risk prediction tools, personalized adaptive interventions, and technologies to address the needs of individuals with disabilities.
Conversational agents engage in two-way dialogue with the user via speech recognition, natural language processing, understanding, and generation. AI/ML enables these agents to assess symptoms, report on health monitoring outputs, and recommend a course of action based on these varied inputs.
The usage of wearables, smart devices, and mobile health applications has grown exponentially. AI can use raw data from accelerometers, gyroscopes, microphones, cameras, and other sensors, including smartphones. ML algorithms can recognize patterns from the raw data inputs and then categorize an individual’s behavior and health status.
These systems allow patients to understand and manage their health and symptoms and share data with medical providers. In addition, as sensors become more ubiquitous in homes, smartphones, and bodies, the data sources for just-in-time adaptive interventions (learning systems that deliver dynamic, personalized treatment to users over time) are likely to continue expanding.
Age is the best predictor of cognitive impairments. Given the aging society, the current care system is unprepared to handle the current, or future load of patient needs or to allow individuals to “age in place” in their homes. Intelligent home monitoring and robotics may eventually use AI to increase independence and improve aging at home by monitoring physical space, falls, and amount of time in bed. Currently available social robots such as PARO, Kabochan, and PePeRe provide companionship and stimulation for dementia patients. Future applications of robotics are being developed to provide hands-on care.
AI Solutions for Clinical Care
Two main areas of opportunity for AI in clinical care are enhancing and optimizing care delivery and improving information management, user experience, and cognitive support in electronic health records.
Prediction, Early Detection, Screening, and Risk Assessment
Clinicians are beginning to access data generated from wearable devices, social media, and public health records; data about consumer spending, grocery purchase nutritional value, and an individual’s exposome. AI will profoundly affect the entire clinical care process with this data, including prevention, early detection, risk/benefit identification, diagnosis, prognosis, and personalized treatment.
Predicting, early detection, and risk assessment for individuals is one of the most fruitful AI applications. For example, diagnostic image recognition can differentiate between benign and malignant melanomas, diagnose retinopathy, identify cartilage lesions within the knee joint, detect lesion-specific ischemia, and predict node status after a positive biopsy for breast cancer. In addition, image recognition techniques can differentiate among competing diagnoses, assist in screening patients, and guide clinicians in radiotherapy and surgery planning.
AI techniques will mainly be assistive, sorting and prioritizing images for immediate review, highlighting missed findings, and classifying simple results so humans can spend more time on complex cases.
AI in Surgery
AI can bring together diverse sources of information, including patient risk factors, anatomic information, natural disease history, patient values, and cost, to help physicians and patients make better predictions regarding the consequences of surgical decisions. In addition to planning and decision-making, AI can be applied to change surgical techniques. For example, remote-controlled robotic surgery has been shown to improve the safety of interventions where clinicians are exposed to high doses of ionizing radiation and make surgery possible in anatomic locations hard to access.
Improving Information Management, User Experience, and Cognitive Support in Electronic Health Records
AI can support and also disrupt clinical decision support (CDS) modalities by enabling a wide range of innovations with the potential to disrupt patient care. For example, mainstream medical knowledge resources use machine learning algorithms to rank search results. These retrieval tools can be linked with the conversational systems that allow clinicians to retrieve patient information from the EHR.
Through information extraction, NLP, automatic summarization, and deep learning, AI has the potential to transform static narrative articles into patient-specific, interactive visualizations of clinical evidence. Furthermore, these visualizations can be seamlessly embedded within the EHR.
Future Healthcare Imaging infra development:
The below picture puts things in a good perspective, summarizing the clinical needs or Wishlist and the science development need. Despite all the technological developments in the last decade, current clinical imaging capabilities are limited to the left box; in other words, most clinical imaging technologies can visualize cancers when they approach 1 cm3.
Our clinical investigations are Biopsy driven; we are still far away from peeking into living cells or in-vivo probes. The ability to observe our cells live and understand them better will be something no less than the god-like capabilities to help ourselves.
Source: https://www.pnas.org/doi/pdf/10.1073/pnas.1508524112
The future developments can be listed similarly to my last blog, along the lines of clinical imaging systems, research imaging, molecular and cellular markers and imaging, imaging physiology, and immune cell imaging. Each has a distinct focus and offers exciting future capabilities.
Clinical Imaging Systems.
Imaging systems can be grouped according to energy type (X-rays, positrons, photons, or sound waves), spatial resolution (e.g., macroscopic or microscopic), or obtained information type (anatomical, physiological, cellular, or molecular). In addition, the increasing sophistication of imaging technologies has resulted in new ways to understand how the energy-matter interaction generates different contrast mechanisms (e.g., magnetic relaxivity, susceptibility, diffusion, temperature, elasticity, electrical impedance, radiation absorption, scattering, and fluorescence).
The current imaging technologies have evolved a lot; we are hurtling towards microscopy performed in live subjects and imaging at extreme resolutions in live cells.
Research Imaging.
In the research setting, live imaging in animals has led to cancer biology, immunology, and brain function discovery. Research microscopy evolved to use confocal or multiphoton scopes with long working distance objectives, special lasers, and unique motion compensation techniques. In addition, fluorescent proteins and exponentially expanded computational power have enabled research contributions.
Future holds developments in automated image analysis, data mining, integrating complex datasets into multiscale models, and developing new visualization tools.
Imaging Molecular and Cellular Biomarkers:
The biomarkers are most useful for in-vivo imaging, studying organs not readily biopsied like in the brain, detecting early cancers, and mapping disease severity and location. Developing multiplexed imaging and cytometry approaches will likely play an essential role in defining new imaging targets.
Finding Smaller Cancers. Most cancers are curable when detected early (>90% in stage 1). Most clinical imaging technologies can visualize cancers when they approach 1 cm3. With the recent advances in image resolution and chemical agents, this size boundary is being pushed toward 1mm3. Hopefully, even smaller-sized cancer lesions will likely be detectable in the future.
To detect minute cancer lesions, tools to determine the properties of precancerous lesions that predict the progression to malignant disease are needed. Some exciting possibilities are expected from new imaging technologies, sensors, and agents through engineering advances, combining blood biomarkers with imaging and developing microscopic imaging tools that can be used intraoperatively or during minimally invasive procedures (i.e., micro-endoscopy).
For example, the first clinical trials from intraoperative imaging with fluorescent affinity ligands or antibodies are promising. These approaches will ultimately change cancer surgery standards by giving surgeons real-time feedback about tumor margins and whether any cancer remains.
Imaging Physiology. The only way to learn about many physiological processes is to watch them in vivo. Critical aspects of complex physiology and disease processes co-occur, including those that seem to conflict. Non-invasive imaging, even clinical imaging, will likely adopt the advantages of spectrally resolving several targets. Integrating comprehensive imaging data can provide unprecedented insight into pathology.
Immune Cell Imaging. Single-cell immunocyte imaging has tremendous potential for deciphering cells’ in vivo spatial distribution, dynamics, lineage, and behavior in disease. High-resolution imaging has recently led to surprising discoveries.
Newer approaches to cell labeling, reporter gene strategies, immune cell imaging, and checkpoint blockade imaging are being explored clinically.
Taking cancer as an example, some of the puzzles being explored are why do some patients respond much better than others to emerging immune checkpoint blockades or why do some patients experience extraordinary toxicities with immunotherapies whereas others do not? Translating the exploration and insights from mouse to man would likely be impossible without imaging, which, unlike biopsies, can sample the entire human body noninvasively.
The Future
The new tools and understanding will ultimately allow new types of measurements. The most valuable techniques will quantitatively and comprehensively access the cellular and subcellular/molecular levels in vivo.
Referring to the figure above, we can hope to resolve the below challenges sooner than we think:
- How do detect cancer early with lesions < 1 mm3 ?
- Develop single-cell imaging deeper than ∼200–500 μm (intravital live imaging through an imaging window implanted into tissue).
- Characterize individual cells’ functional states within tissues and tumors?
- We need vastly increased data acquisition speeds to accelerate imaging; how do we improve the spatial resolution by 10- to 100-fold without increasing acquisition times?
- How can we push multiplexing: i.e., simultaneously imaging 10–100 targets?
- Develop chemical tools: like brighter, small footprint fluorochromes that are biocompatible and can be used as sensors
- Develop a complete set of mouse models with genetically encoded fluorescent proteins in all relevant classes of immune cells in addition to lineage tracers for each of these cells.
- Accelerate the development of human microscopic imaging through endoscopes and probes.
- Develop miniaturized sensors and implantable microscopes for long-term imaging on the scale of days to months.
Within a couple of decades, we will push boundaries to make diagnosis way less invasive and zillion times sharper, more preventive than post-disease, and at the same time cheaper. It might still be a century away from finally articulating what quantum computing means and how we can develop the probabilistic computing constructs, but that is an even more significant change to build on these capabilities.
Disclaimer:
I currently work full-time at Swiss Re, Bengaluru. The blogs and articles on this website www.balajos.com are the personal posts of myself, Balachandra Joshi, and only contain my personal views, thoughts, and opinions. It is not endorsed by Swiss Re (or any of my formal employers), nor does it constitute any official communication of Swiss Re.
Also, please note that the opinions, views, comprehensions, impressions, deductions, etc., are my takes on the vast resources I am lucky to have encountered. No individuals or entities, including the Indian Institute of Science and NSE Talent Sprint who have shown me where to research, or the actuarial professional bodies that provide me continuous professional growth support, are responsible for any of these views; and these musings do not by any stretch of imagination represent their official stands; and they may not subscribe/support/confirm any of these views and hence can be held liable in any vicarious way. All the information in the public space is shared to share the knowledge without any commercial advantages.