5 Types of Tunnel Vision, AI Tech Giants Overlook to Grab Low Hanging Fruit in AI BioMed
In the history of artificial intelligence, an AI winter is a period of reduced funding and interest in artificial intelligence research. A marked increase in AI funding, development, deployment, and commercial use leads to the idea of the AI winter being long over.
Concerns however, are occasionally raised that a new AI winter could be triggered by overly ambitious or unrealistic promises by prominent AI scientists or overpromising on the part of commercial vendors.
I will give you a glimpse of where AI Tech Giants demonstrated a certain blindness towards abundant low hanging fruit problems easily to be solved on a global scale. But AI Tech Giants should be warned that “they only have one shot, or one opportunity to seize everything they ever wanted. In one moment will they capture it or let it slip”?
Have you ever observed the way giraffes eat as browsers? It’s surprising how similar giraffes behave and the way AI Tech Giants behave in the medical arena these days.
As the giraffe is reaching out to find the highest leaves, he is not able to see, smell, taste or just be aware of the low hanging fruit a few inches above the ground.
Constitutionally, anatomically, the giraffe is not able to act differently, even more, if he comes close to the surface for ingesting water, at that very moment a kill is never been closer for the giraffe than at any other time.
- AI Tech Giants act like Giraffes.
Well AI Tech Giants act similarly, exactly like the giraffes! Without any doubt, the board members, stakeholders and engineers of the AI Tech Giants use their full cognitive spectrum of creativity to try to understand and commercialize solutions for the medical community by promising solutions for technological problems that need to be urgently solved.
The low hanging fruit here, is to be considered, any action a healthcare professional takes at least a hundreds times a day in a way the same professional gets frustrated why simple steps are still not automated or at least more structured and accessible.
Take the following example:
Medication and its prescription, considered as a process, or a task if you will, should be quite straightforward to perfectly integrate the act of prescribing in the clinical flow as the majority of the data and metadata of each pill, tablet, IV injection etc consists of at least the following elements.
- a generic name or active substance name; Ibuprofen
- a commercial name; Ibuprofen
- a pharmaceutical form / how supplied; tablets/ oral intake
- a mass or volume; 200 mg
-a dosage: 4x1 tablet /24 h
With the US Pharmacopeia 2020 PDF version requiring less than 15 MB memory of content, Real World Voice (different languages, different dialects, etc.) to Structured Text should not be that difficult to develop, run as proof of concept and deploy this universally, as this process is done hundreds of times a day, by doctors, nurses, nurse practitioners, pharmacists etc.
No other industry like BIOMED would not have solved this decades ago, why BIOMED is not aware of a solution, healthcare providers are crying for decades, is completely unclear and hundreds of similar examples could be mentioned here.
Other examples falling under this subheading are the false alarms. I agree, we are investing huge amounts of energy to develop early warning systems which are both valuable but probably not universally applicable. False alarms of the type asystole alarm with one missing ECG electrode, a normal plethysmography curve and normal hemodynamic blood pressure.
2. Healthcare is a completely different ball game!
I get the impression that the Artificial Intelligent ball game is played with only referees on the field, a lot of supporters and with players at the stage.
As such, the game is not played, Why?
Healthcare is such a different industrial branch. There is a dichotomia: “or you are fully immersed into it and understand who the key players, the influencers, the invokers, etc are” or “your added value is minimal without on the spot deep training”.
In this sense, digital transformation is not a simple change. It is not just a question of the modernization of medical infrastructures or technologies. It is a continuous and complex process, multidimensional, linked to social, economic and technological factors that transcend the walls of hospitals. It implies a change in the mentality of the whole process of care, placing the patient at the centre of this transformation.
Even if retrospective studies are well defined (in relation to samples and features) and if the medical community is confident with models and results, successful predictive analytics and application of cutting-edge machine learning algorithms often demands substantial programming skills in different languages(eg, Python or R). This migrates modeling from the domain expert to the data scientist, often missing the necessary domain expertise, and vice versa. Additionally, data analyses are highly creative processes and there are no detailed recommendations for conducting such research. High-level steps for conducting this research is described by the cross-industry standard process for data mining, which is breaking down the life cycle of an analytics project into six phases: business understanding, data understanding, data preparation, modeling, evaluation, and deployment. However, specifications of each problem prevent the development of a standardized analytics process on an operational level. This ultimately leads to the slow development, adoption, and exploitation of highly accurate predictive models, in particular in medical practice, where errors have significant consequences (both human and financial).Obviously, a close and continuous collaboration between domain experts and data scientists would solve this problem, but this is not always feasible. Many efforts have attempted to overcome this problem in recent research. One of the directions is formalization of domain knowledge through medical ontologies and integration with data-driven models . This approach aims to allow for data-knowledge fusion and to reduce the need for additional specialization of domain experts in data science and vice versa. Another approach is development of visual analytics tools that enable a faster learning curve and powerful analytics that can be conducted by domain experts.
As such we dive into the most important, most abundant error AI Tech Giants make:
3. Get Doctors On Board! No soccer game was won with only referees!
Getting doctors, your user group on board as soon as possible in as much as departments and roles as possible, is THE only way to adapt technological advancements to what is really needed on the work floor. As in the earliest years , some institutions had statisticians on the floor. Dr. Anthony Chang from AiMed io works on a daily basis with AI specialists on the floor. As the doctors need to learn the language of developers and company goals, the company and engineers also need to fully understand how medicine is complex in its own way. As such the gap between the domain experts (data science vs medical) will be reduced and could evolve to inter and trans-disciplinary team effort, rather than working and scoring in silos.
4. Forget Interview Questions Without Medical Context
I recently learned that an extensive interview with an AI Tech Giant gained more traction than the classic calculus, probability, machine learning quizzes. I never understood how a quiz-like interrogation which more frequently was taken from libraries of questionnaires could select the better candidate in a first selection. It reminds me of the unsolved link of the open vacancies on Linkedin and the candidates waiting with their diploma’s, credits, etc interested in exactly the job as described in the company open vacancy announcement. With unsolved I mean the complete lack of a solid talent offered — talent needed solution, which based on the content available from any media related to a potential candidate, sufficient information should be available to get a headhunting process on the rails.
5. You only have one shot, one opportunity
Today there is no industrial counterpart where you could learn the appropriate skills, fake or try to infiltrate into the medical arena without being caught and digested if you even smell like a non-professional. Therefore all topics above have a reason why they were written down.
