Image for post
Image for post

Safety and efficacy are of paramount importance in the development of medical devices. To its credit the U.S. Food and Drug Administration (FDA) has spent the last few decades developing and refining a highly structured, evolving, and largely consistent framework for medical device evaluation. The central anchor of this framework are the User Needs including the Intended Use and the Indications for Use. The Intended Use is what the medical device is used for; and the Indications for Use are the set of reasons, circumstances, and environment for which the device is used.

Medical devices are never approved by the FDA for nonspecific usage. Instead, they are approved for use in a highly specific context, clinical setting, demographic, user group, hardware accessory, clinical workflow, etc. These attributes together are the Indications for Use of the device and they define the scope within which the FDA deems the device to be safe and effective. As such, any academic exercise that sets out to evaluate the safety and efficacy of an approved device must do so within the device’s Indications of Use (or reasonably foreseeable misuse). This is especially important in Artificial Intelligence, an area of rapidly expanding importance and one whose many stakeholders need to be properly informed. …


This article focuses on the American healthcare system, and uses diabetes management — and diabetic retinopathy in particular — as an example to explain the value-based care landscape.

Image for post
Image for post

Value-based healthcare incentivizes quality and resourcefulness in care. Both AI and Telemedicine present an opportunity to swing the pendulum towards higher quality, more affordable, and more accessible care. However, several stakeholders need to be properly aligned for progress at pace. On December 2nd 2020, the Center for Medicare & medicaid Services (CMS) released its final rule for CY 2021 Physician Fee Schedule. …


Image for post
Image for post
CT Angiogram of the Head and Neck

As I was seeing patients in the retina clinic this morning, we had gotten off to a great start. I had done a few intravitreal injections and one pan-retinal laser photocoagulation procedure. At that point, I got informed that one of my patients whom I was yet to see had developed symptoms of an acute stroke. I rushed into the room. Following a very quick assessment, I asked that the patient be immediately transferred to the Emergency Department for a stroke protocol assessment. The rapidness of my decision was based on the stroke management mantra “time is brain.”

From the onset of a stroke, one has approximately 4 hours ideally to administer a certain “clot bursting” drug called tissue-type plasminogen activator (tPA) and or physically remove the clot via a procedure called mechanical thrombectomy — each of which can improve outcomes. And typically, only interventional neuroradiologists have the proper training to perform either of these interventions. …


Image for post
Image for post
Cayley Graph of Quaternion Group. By Nielmo (License)

Mathematics is dangerous. John Carlos Baez, a Theoretical Physicist at U. C. Riverside and an excellent science communicator, tweeted about the 5/8 theorem a few days ago. Reading his tweet, I was hit by a related observation that the commutativity expectation of the quaternion group equals the number of conjugacy classes divided by the order of group. Additionally, I ‘felt’ that Hamiltonian groups must be 5/8 maximal. I felt so, because Hamiltonian groups are non-abelian Dedekind groups. In other words, despite being non-abelian, they possess a high degree of abelian-ness in that every subgroup commutes with every element of the group. Not being active in the Group theory research community, I was not sure if my observation was novel or not. What do I know? I am just a medical doctor. Nonetheless my observations and conjecture where certainly interesting to me, and I was curious to know if they are true, and more importantly if they generalized. Thus began my quest. By the end of the weekend I had named the theorem and had derived a complete original proof of it. I subsequently surmised that the theorem was almost certainly already known to be true, even though I could only find one source that alluded to it; and that source provided no accompanying proof. I learned a lot from the endeavor and drew up some future work direction for someone else. I have patients to see. …


Image for post
Image for post
Kubernetes, Letsencrypt, and Google Cloud Logos (Public Domain)

Security and identifiability remain of paramount importance as we mold and embrace the exciting cloud-centric paradigm shift occurring in enterprise software development. As the landscape shifts from monolith towards microservices (and back), from on-premise computing to the cloud, and from adhoc provisioning and configuration to infrastructure-as-code, the encryption and domain name configuration infrastructure are also changing. Therefore in this tutorial, we will specifically cover the following two topics:

  • Configuring Cloud Domain Name Services (DNS) in Google Kubernetes Engine (GKE) with custom domain names
  • TLS encryption of network traffic between our cluster and the internet

A pre-requisite for this tutorial is the content covered in my prior tutorial on GKE ingress mechanism. Further, we assume the following infrastructure is available and…


Image for post
Image for post
Credit: Phys_org. CC0 Public Domain

Due to the ongoing COVID19 pandemic, much of the world has implemented social distancing measures, and video conferencing has become a necessary means of continuing business during these trying times. For example, Telehealth and remote AI diagnostic systems have suddenly taken center stage in healthcare. However, of late there has been a spate of security breaches in video conferences around the world. And though most have been pedestrian scrapes of video conference URLs from social media, others supposedly have been more sophisticated; including breaks into the backend of video conferencing service clusters. One way for vendors to mitigate against such security vulnerabilities is to decrease the attack surface of their microservice clusters. In particular, by ensuring that one’s cluster is only accessible through a single gateway, one significantly decreases the chances of a malicious agent gaining access into the cluster. This is in contrast to clusters with multiple potential entry points which can be exploited by bad actors. …


Image for post
Image for post

As the world grapples with Coronavirus and seeks a well-calibrated effective response to the pandemic, there is increasing recognition of the important roles Artificial Intelligence and TeleHealth will play in building resilient healthcare systems that work. To be useful, machine learning algorithms have to be deployed into the real world in a form that is accessible, reliable, secure, and that makes a positive impact in our daily lives. To accomplish these goals, a number of infrastructural components have to be firmly in place. Specifically, our machine learning models need to be deployed into an ecosystem or network of other services which depend on them and vice versa. …


Image for post
Image for post

There has recently been great controversy on social media over comments made on Feb 20th 2020 by Geoffrey Hinton, who by consensus is considered the most senior distinguished AI researcher in the world, and who is a recent winner of the Turing Award (the “Nobel Prize” of computer science). Here are Hinton’s comments:

“Suppose you have cancer and you have to choose between a black box AI surgeon that cannot explain how it works but has a 90% cure rate and a human surgeon with an 80% cure rate. Do you want the AI surgeon to be illegal?” …


Image for post
Image for post
VAE Illustration by Stephen G. Odaibo, M.D.

When I was in graduate school in computer science at Duke~2007/2008, the then DGS of statistics (Merlise Clyde, I believe, now Chair) attempted to recruit me to leave the computer science department and join the statistics department. I had an emerging appreciation for statistics, but it was not yet fully formed. Her offer surprised me and sounded interesting but I declined it. Had I foreseen how ML and Statistics would merge today*, perhaps I may have obliged. But the Spirit of the LORD had not led me there at the time. Stats is superbly interesting and so much fun now. But CS is also still lots of fun. So who knows? Maybe I should have “quadruple majored”. I digress. …


Image for post
Image for post
A Simple Attention Mechanism

Attention mechanisms are very important in the sense that they are ubiquitous and are a necessary component of neural machine learning systems. However, for that very same reason, they are not so special. Contrary to what their newfound recognition would suggest, they are not some exotic new device only present in certain “next generation” models. Instead, they are a component of essentially all deep learning models. Giving them an explicit name and identity is certainly useful, but has misrepresented them as more special than they are. Attention is inherent to neural networks, and is furthermore a defining characteristic of such systems. …

About

Dr Stephen Odaibo

Physician. Retina Specialist. Computer Scientist. Mathematician. Full Stack AI Engineer. Christian. Husband. Dad. CEO/Founder RETINA-AI Health, Inc.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store