How I Fail S02E11: Pradeep Reddy Raamana (PhD’14, Medical Imaging and Machine Learning)

Dr. Pradeep Reddy Raamana is an Assistant Professor at the University of Pittsburgh. He is interested in developing multimodal biomarkers for brain disorders, and the necessary data science tools to realize personalized medicine. He founded the special interest group on neuroimaging quality control (niQC) at the International Neuroinformatics Coordinating Facility. He is a passionate advocate for quality, reproducible and open science. He leads the Open MINDS lab @ Pitt (now hiring!), where they develop multidisciplinary techniques for computer-aided diagnosis (CAD) and precision medicine, with a motivation to improve data quality, reproducibility, performance of predictive models and their potential for clinical translation.

Hi Pradeep, thanks for joining How I Fail! Next to your official bio, could you say a bit more about yourself?

Thanks for inviting me, Veronika! I would like to think I am a simple straightforward guy trying to improve the world around me in various ways I can. I think a good way to be happy is to make others happy! They sound cliché but that is in fact where I derive my motivation to find value in my actions, work and life in general.

That is also a key motivation for me to choose the academic career, which I believe would give me the opportunity to 1) develop necessary techniques and tools to improve healthcare, and 2) [at some point soon] provide a sufficiently stable and independent platform to speak my mind and pursue solving whichever problems (broadly speaking) I believe would make the most impact. I hope to pursue it by rejecting the traditional careerist approach, and help reform it by eliminating various sources of toxicity in academic life for all stakeholders involved including trainees and faculty.

A recent step I am taking in that direction is to “boycott” all interactions (reviewing, editing and submitting any academic outputs including papers and books) with for-profit publishing businesses like Elsevier. The motivation is to eliminate the subjective (and arguably disproven) concepts of “prestigious” and “high-impact” journals (by relying on high rejection rates and impact factors etc) from our peer-review discourse as well as in hiring and promotion policies.

I believe sticking to publishing only in society-run journals that are accountable to the members of their scientific society would give us the power to choose the peer review criteria (focus on validity, not on accumulating citations to charge higher subscription fees), and optimize them towards what is ethical and best for science. This would also reduce the costs for publication and be more inclusive.

Given great adoption of open science we are seeing which includes transparency in peer review, we can adapt the policies of a society-run journal based on the best available evidence. Within a society-run journal, members would have option to reject many unethical and toxic practices we notice in many for-profit journal businesses such as

  1. picking for “shiny” topics and encouraging authors make inflated claims,
  2. extracting within-same-journal citations from authors (to manipulate their IF),
  3. unnecessarily large number of revisions to leech off many-fold submission fees as well as to
  4. create a false appearance of faster processing time from submission to publication.

I urge everyone to join me, esp. those with sufficient job security. 

What’s a memorable failure for you?

Great question! Besides the typical failures we all face, I think the most memorable for me would go back all the way to the first year of my PhD where I felt the euphoria of building a “perfect biomarker” to diagnose AD “accurately”. Like 100% accuracy in my evaluation (using an SVM classifier I think – they were all the rage then!). I thought I won the jackpot and was super happy. That lasted only 10 minutes or so until I discovered I was totally double dipping!

On Twitter you shared about null results during your postdoc – can you explain more about this?

I was referring to a key project in my postdoc to build neuroimaging “biomarkers” (loosely speaking) for Major Depressive Disorder (MDD). When I was just getting started on it, there were few review papers implying the possibility and potential of neuroimaging to predict response to treatment.  I did have sufficient expertise with both machine learning and imaging-derived features: the focus of my PhD thesis was exactly that but focused on Alzheimer’s disease (AD) and related neurological disorders.

I picked it up with enthusiasm but after a long series of attempts to evaluate the predictive utility of know biomarker candidates as well as some new ideas we had, I realized neuroimaging-based features (various versions of morphometric and functional connectivity) were no better or only slightly than chance. As you know it’s harder or impossible to publish null results, except in specialized/non-popular venues. Hence, I kept digging deeper and wider to try find a “somewhat positive” result, and I must say I failed thoroughly. However, I must say I succeeded in thoroughly understanding various challenges related to predicting response to treatment in MDD well.

If you ask me how I viewed it differently then and now: then, I saw it as an exciting, important, high-risk high-reward challenge, that will demonstrate my expertise and diversify research “portfolio” so to say. I was skeptical of the published results (based on my previous understanding of the AD literature that most biomarker studies reported inflated and/or overfitted results).

But looking back I feel I didn’t think sufficiently through various possible outcomes and whether they align with my career plan i.e. I should have asked the hard questions first: 1) will the results be publishable, and 2) do they add to and merge as a coherent part of my research program in my job talk etc? However, I don’t regret that effort at all! I learnt a lot from this exercise that I wouldn’t have learnt otherwise esp. challenges in in the world of psychiatric disorders. But I am thoroughly disappointed in the system, structure and community around us that would rate that effort (without big papers to show) towards nothing.

Is there a story behind your name “Future Distinguished Emeritus Prof” and bio “without an agenda”? 🙂

LOL, that title is a joke on the high prevalence of Named Professors / Chairs in the academy, which I believe are related to donations into the university. Given the already-hierarchical nature of the academy and concentration of power into few rich clubs, this never-ending quest for more titles incentivizes “generation of money” from new sources proposing all sorts of research. This quest to climb the ladder higher and higher leads to piles of papers in flimsy fields that don’t really move the needle in the context of larger challenges to society, that they are supposed to help with.

This is also related to some traditional biography pages I read which are just a really long list of such “titles”, and I remember saying to myself “OK.. that’s a stack of “trophies”, but tell me what you actually do?”. I personally believe titles shouldn’t get in the way of a good conversation or a debate, and we must evaluate comments on their substance and merit. So, the prefix is just to indicate I will also have few titles in the future. Some on Twitter told me it made them chuckle! J

The “without an agenda” in my Twitter profile is to clarify how I try to conduct myself there: to keep it real, unfiltered and without a specific agenda towards promoting a certain set of topics or people. Few years after I started using Twitter routinely, I’ve noticed how some powerful people promote only those (people or papers) that suit their agenda, and not all the good work that exists. That is really upsetting when it is done by those claiming to be a champion of open science. This saddened me deeply, and I decided to try keep it real and be myself.

The “without an agenda” is also a disclaimer and “warning” of sort for people thinking of following me (many find me based on my research) on what sort of content/feed they can expect when they follow me. I am not trying to virtue signal here, and I probably have blindspots and biases myself. I hope to learn them and correct them to the extent possible.

It seems that you are pretty open about topics others might be avoiding. How do people usually respond to this?

Great question! I’d love to hear from the community what they think of my timeline 🙂

As a true believer in open science, I think it is essential to be as open and as honest as possible. As is true in broader life (relationships, politics etc), only honest discourse from everyone leads to real and sustainable progress for all stakeholders involved.  So, I try not to filter any comments or questions I have and try providing an open outlet/forum to discuss all topics, regardless of whether I think that particular tweet or topic would be well-received or not. This process of sharing (past some basic filtering) also helps me a keep a journal of sort over time, and over multiple debates on similar topics.

As for how people respond to this, I think there is definitely those who listen to what I have to say, offer thoughtful responses, advice and support in their own ways. I’d like to sincerely thank them, and consider them my friends! They certainly keep me going, as I don’t think I’d have decided to pursue the academic career if I didn’t find that open and supportive community in time.

That said, they are more on the smaller minority, and I feel there is a clear and strong tendency to avoid any “negative” and non-work topics. Engaging in these debates is considered “unproductive” by some. I might be wrong but one of the reasons for this is many researchers treat academic Twitter as another social media channel to promote their work or themselves (“brand management” comes to mind). That is their choice, and that is not wrong, but it doesn’t help with facilitating important open conversations to be had.

There is also a clear and noticeable tendency to popularize who are already popular e. g. an identical idea or comment from a “well-known senior researcher” would get disproportionately more attention than otherwise. I don’t claim to understand how or why that happens, but I think the tweet should matter much more than the tweeter.

Are there any other topics we (collectively) are avoiding, that we need to discuss more often?

Yes – there are a few (which I have been trying to shout out loud and often):

  1. More debate and realization towards how we the academics are squarely the source of many of the fundamental issues in science
  2. More honest communication with prospective PhD students on what academic life is really like, and the bleak prospects for many towards a traditional Professorial career, and why we should strongly discourage many to pursue a PhD in the first place
  3. Why we should eliminate the concepts of “prestige” and ranks in our discourse, starting with journals (Impact Factor etc), and then universities

Is the situation changing? Do you see differences, for example across different fields?

Yes! Based on my experience and bubbles (which is arguably limited and likely not representative of the full academia at large), the “computer sciences” seem to be more open compared to the more “clinical” and life-sciences fields. This may be due to differences in training, and perhaps because the tools and platforms to share (relating to computers and software) are much more natural and integrated to them. I have definitely noticed “hierarchy” and “authority” play a bigger role in the more clinical / “hardware” fields relative to the “software” fields. Again, I could be less than accurate as this was based on my own observations and anecdata.

Is there anything that you are currently failing at yourself?

Yes! I suck at time-management and sticking to my plans and priorities esp. in the medium term.

Do you have a success that is not a traditional one, like jobs or papers, that you are proud of today?

As I participated in the American Elections debates online over the years, I noticed they were only worsening the political divide and weren’t helping with bridging the gap! Hence, I’ve been itching to meet the real American voters in the real world and something “real” instead of another Twitter debate. While the pandemic didn’t really help this cause, I was able to visit many poll centers in the Greater Pittsburgh area to distribute water and food on the historic 2020 American Election Day.

The original idea was to help the voters stay in line and make sure they vote as Pennsylvania was a critical swing state. However, as large numbers voted by mail, the lines weren’t as long as expected. So, I was able to help the poll workers themselves! They were visibly and clearly delighted to see somebody offer them water and chocolates. And that was an experience I wouldn’t forget for sometime.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Mastodon More Mastodon