Innovating with Biomathematics

Photo by National Cancer Institute on Unsplash

Introduction

The medical sciences started out their path as a “holistic” discipline: the information they used to make use of reality was not first-hand, and theories were mainly speculations. In the last centuries, more intensely in the last decades as a result of several technological developments, the medical sciences have been bombarded with a wave of new technologies and paradigm shifts coming from almost all possible directions, especially coming from computer science and mathematics.

On this short assay, we focus on how computer and mathematical endeavors can be used to be innovative on medicine; I am thinking of startups, the current innovative “buzz word”. Since this is a short letter, we just scratch the surface. I have a course on the topic with more details on Udemy[1], for the interested readers.

As a result from a postdoctoral fellowship research period Pires (2021), I have done the talk Innovating with Biomathematics: the challenge of building user-friendly interfaces for computational biology (Thursday Morning Science, 2021). This is a short assay/paper/communication version of the talk, merged with new insights after the talk and the research period.

A considerable amount of thought herein are a distillation of Pires et al (2021), seen from another perspective, more distant from the coding part, which is more technical. Herein, I want to lay down a theory, if I can call this reflection so. This theory can be interesting to computational biologists and web developers to create innovative systems that use models as their inner-working gears; herein, I am using computational biology and computational intelligence as generic terms to related areas, e.g., computational biology encompasses also systems biology, and computational intelligence also encompasses artificial intelligence. Thus, I kindly invite the readers to think all these scientific areas as one, one single piece of endeavor, and the final goal is creating online platforms for supporting medical doctors and biologists, life scientists in general, to solve their daily-professional problems. The most ubiquitous one is decision making, e.g., is this patient sick or no? should I consider this diet or not? would this species vanish or not? would the COVID pandemic spread or not? All those questions are decision making scenarios, which can be aided by models.

Innovation and publications: how publications can hinder innovation

As Spicer and Roulet (2014) bring to attention “as a direct consequence, established journals are usually biased against innovative work.” Additionally, I heard during my postdoc from an innovative researcher in healthcare about the prejudice against people that innovate, which means less publication in the traditional sense; mentioning it since I just published a book on the topic (Pires, 2022), on how Publish or Perish (POP) has created several issues in science, one of the them is regarding innovation, taking knowledge from science to society. Innovating with biomathematics is all about how we can use our models from computational biology (e.g., mathematical biology) to support real-scenarios, instead of just publishing them, creating a dead end for innovation. Sometimes, I wonder how less Google™ would have done if they were concerned about publications. As one example, I do believe that TensorFlow.js Nielsen et al (2020), the Google’s version of the classic TensorFlow Rundo and Tangherloni (2017), used in deep learning, can be quite imperative in innovating with biomathematics. A considerable amount of works is done in medicine are using artificial intelligence, see my postdoc report (Pires, 2018).

Risk of innovation vs. risk of market: how it can impact innovating with biomathematics?

Blank and Dorf (2012) organize startup endeavors in three groups, when it comes to risk: i) innovation; ii) customer/market; iii) both innovation and customer/market. Innovating will biomathematics will fall into the 3rd situation.

Discussion on Academia Edu

From one side, we have the risk of innovation since computational biology in general is based on ongoing researches, both from the biotechnological and from the computational biology side (e.g., precision medicine and targeted therapy). Once one gets the system working accordingly, one may have the risk on costumer/market: the usage of mathematical models in medicine. One point that I like to defend on this scenario: do not wait for the model to be perfect, startup entrepreneurs, they use the market as compass, called ‘pivoting’, we could do so as well when innovating with biomathematics.

The culture of using computational biology models on real-world scenarios is an ongoing branch; platforms like the one proposed by Pires et al (2021) hope to support it accelerating by offering easy-to-use platforms, that can be used to support on weighing the medical diagnosis decisions (Figure 1).

Figure 1. Going from models to innovation.

According to online discussions with two experts on the field, namely (Kolodkin, 2021; Karr, 2021), some challenges those models may face, which can be classified as innovation risk, and even market risk at some level: i) the models currently cannot represent the reality, as so (Karr, 2021) is extending his famous and respected model for including human cells; Kolodkin, (2021) discusses a possible, I would say a distant, future of virtual humans. Thus, the innovation risk here is the fact that even though on academia those models are widely used, on real world they are quite limited for explaining complexity. In part, this is because they depend on medicine and biology: recently, the Nobel prize was to David Julius and Ardem Patapoutian “for their discoveries of receptors for temperature and touch”[1]; imagine someone modeling this system, now they have to revise their model, for the better.

How much do we have to know to predict? do not mistake understanding with prediction! The illusion of validity[2]

Innovating with biomathematics is all about using models to assist on decision making.

Are models reliable?

We have essentially two big models when it comes to computational biology: black-box vs. white-box models. They are, respectively, top-down vs. bottom-up approaches. They differ essentially on what they use mainly as main ingredient: the former uses data, lots of data; the latter, theory, “first-principles”.

Example of black-box models are artificial neural networks, used on machine learning, e.g., deep learning; example of white-box are differential equations. When it comes to human-like decisions, black-models may be the best. For instance, they can be easily used to decide on data-based if a patient is sick; whereas white-box models are mainly for understanding a system, simulating their behavior. We may say that black-box models are the closest one to be applied in innovation. Machine learning is everywhere, see Nielsen et al (2020) for very interesting applications of deep learning on several problems outside medicine; I have some examples on my postdoctoral report for medicine Pires (2018).

Sunstein et al (2021) presents some intriguing insights about models, including medicine: “In one of the three samples, 77% of the ten thousand randomly weighted linear models did better than the human experts. In the other two samples, 100% of the random models outperformed the humans.”. For medical context, the interesting of innovating with biomathematics, they cite Meehl (2013), which presents several cases where models were able to outperform humans: they call it mechanical vs. clinical judgments.

How much does it cost?

One of the pillars behind my assertion that the time is ripe to innovate with biomathematics is how much we have now for free. On Pires et al (2021), I have built in collaboration a full platform for computational biology, back and frontend without any cost: the codes are MIT license. And it is not an exception. Even though Matlab® and Mathematica® are paid, and quite powerful, we have alternatives for free, e.g., Octave and Scilab. For the frontend, as an example, mathjs is free, and do a lot, even symbolic calculation Pires (2021b); even Mathematica® has a set of free open access packages for web applications Pires (2021b).

Nowadays we have what I like to call “the culture of free for small”: you can use for free, should you want to scale up, you pay as you need, on demand. This is perfect for innovating.

Advantages of web applications for innovating

are we still desktop applications driven? I think not” (Pires, 2021c).

Without being detailed, strong suits of web applications, compared to desktop: i) Simplicity; ii) Aesthetic; iii) Updatable; iv) Efficient; v) Removable; vi) Supported.

Final remarks

As final remarks, based on my researches, we have a time ripe for mathematical and computational models in medicine. I do believe that innovation may be the best path, compared to simple academic studies; the academics either shift to innovate or some company such as Google™ may take the lead, as they have done in several areas such as self-driving cars. I hope we academics would wake up and do not let this important task to others, since most of the hard work we have already done, the last steps are relatively simple.

On this talk, it became evident to me how important are those models.

References

Blank S, Dorf B. (2012). The Startup Owner’s Manual: The Step-By-Step Guide for Building a Great Company. K & S Ranch.

Kolodkin, A. Virtual Human | in conversation with Alexey Kolodkin. [YouTube live]. 2021. Accessed on 27/04/22: https://bit.ly/3vnxs9H

Karr, J. Whole-cell model | in conversation with Jonathan Karr. [YouTube live]. 2021. Accessed on 27/04/22: https://bit.ly/3LsazYj

Nielsen, ED.; Cai, S; Bileschi, S. Deep Learning with JavaScript: Neural Networks in TensorFlow.js. Manning; 1st edição (11 fevereiro 2020).

Meehl, CE. linical Versus Statistical Prediction: A Theoretical Analysis and a Review of the Evidence. Echo Point Books & Media (12 fevereiro 2013).

Pires, JG. Semantic biomedical image segmentation. Federal University of Bahia, Graduate Mechatronics program. postdoctoral report. Salvador: Brasil, 2018. https://bit.ly/3rVPduC

Pires et al. Galaxy and MEAN Stack to Create a User-Friendly Workflow for the Rational Optimization of Cancer Chemotherapy. TECHNOLOGY AND CODE. Omics Technologies Toward Systems Biology. Front. Genet., 18 February 2021. https://doi.org/10.3389/fgene.2021.624259

Pires, JG. Desenho e Desenvolvimento de uma aplicação web baseado em JavaScript para a plataforma Teranóstico. postdoctorial report. CENTRO DE DESENVOLVIMENTO TECNOLÓGICO EM SAÚDE: Oswaldo Cruz Foundation. 2021. https://bit.ly/3vkAtYm

Pires, JG. Angular and Numerical Analysis: creating a library for numerical integration of ODEs, Euler’s method. Published in Geek Culture. 2021b. Accessed on 28/04/22. https://bit.ly/3kiNkUu

Pires, JG. My selected assays from Medium on Computer programming: Angular, JavaScript, Machine Learning, TensorFlow.js and more! Vol 1. (My writtings on Medium) (p. 87). Edição do Kindle. 2021c. https://amzn.to/3xZYq9c

Pires, JG. Qual o real papel do revisor acadêmico?: Como jogar uma espécie de loteria, propensa a polarização e facilmente abusada. Edição do Kindle. Self-published, Ouro Preto: Brasil, 2022.

Sunstein, CR.; Kahneman, D; Sibony, O. A Flaw in Human Judgment. Little, Brown Spark (18 maio 2021).

Spicer, A; Roulet, T. Hate the peer-review process? Einstein did too. June 2, 2014. Retrieved April 23, 2022, from https://bit.ly/3MFMygB

Rundo,L; Tangherloni, A. Biomedical Image Segmentation and Analysis using ML and CI Techniques. 1st CIBio-CBIC, 2017. [YouTube]. https://www.youtube.com/watch?v=1oTWxxe9YJM&t=89s

Thursday Morning Science — 08/04/2021. (2021, April 8). [Video]. YouTube. Accessed on 28/04/22. https://www.youtube.com/watch?v=iJSLntPDLcU&t=6s. University of L’Aquila. Title: Biomathematics: the challenge of building user-friendly interfaces for computational biology.

[1] Innovating with Biomathematics: The challenge of building user-friendly interfaces for computational biology. https://bit.ly/39coMds

[1] https://www.nobelprize.org/prizes/medicine/2021/press-release/

[2] Based on Pires, JG. How much do we have to know to predict? do not mistake understanding with prediction! The illusion of validity. Published in DataDrivenInvestor, 2021. https://bit.ly/3vS0kFR

Academia Letters, June 2022 ©2022 by the authors — Open Access — Distributed under CC BY 4.0

Corresponding Author: Jorge Guerra Pires, jorgeguerrabrazil@gmail.com

Cite as

Citation: Pires, J.G. (2022). Innovating with Biomathematics: the challenge of building user-friendly interfaces for computational biology. Academia Letters, Article 5792. https://doi.org/10.20935/AL5792

Upcoming live discussion, join us!

--

--

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
Jorge Guerra Pires, PhD

Jorge Guerra Pires, PhD

Independent Researcher and writer at Amazon. Visit my profile on Amazon: amazon.com/author/jorgeguerrapiresphd | “I want thinkers, not followers!”