The healthcare industry is one of the most striking beneficiaries of data sciences. In post-Covid world medical diagnostics, medical treatment is becoming more efficient and accessible, medical treatment more personalized, and medical research more data driven.
Data scientists drive innovation across the healthcare sector; like chatbots that can help patients find a good physician, productivity applications that can automate administrative tasks, and recommendation services that can identify patients who could benefit from a new clinical trial.
Dr. Bushra Anjum is among the few inspiring ladies leading in data sciences and is currently based in California. Born and raised in a middle-class family in Pakistan, her father served in the Pakistan Army, and she’s the youngest of three daughters. While doing her master’s in computer science from LUMS, she showed interest in studying abroad for a Ph.D. degree. Fifteen years ago, she came to the US, and with her hard work and devotion to her career, she’s now a trained data scientist, having worked extensively with predictive analytics. Currently, she is working as the Senior Analytics Manager at Doximity, a San Francisco-based health tech company.
Below is the recent conversation we had with her. Dr. Anjum discusses the promise of data in the field of healthcare, some fascinating details about the data products she has been building, her volunteer ventures and how women are essential to create an equitable data driven future for all of us.
Q: Let us know about your life and career? Who inspires you the most for an offbeat career like a health data scientist?
I am a trained data scientist, having worked extensively with predictive analytics, and currently working as the Senior Analytics Manager at Doximity, a San Francisco-based health tech company. I received my Ph.D. in Computer Science at North Carolina State University in 2012, then served in academia (both in Pakistan and the USA) for a few years before joining the tech industry. I joined Amazon and worked for the Prime team, where I was a backend engineer for four years. My research background is in performance evaluation and queuing theory. Combining that with the engineering expertise gained during my tenure at Amazon, I switched to the field of data science, which brought me to Doximity. I joined as a data scientist and later got promoted to leading the revenue wing of the company.
The inspiration for my work is the patients and their caregivers, family, and friends who are not hired to do the job but do it out of love and concern. Our health care systems worldwide need to be effective, yes, but equally if not more important, they need to be empathetic. No one needs software complexity, vague instructions, incorrect diagnostics, and unsafe personal and financial data on any given day. However, consider dealing with all this on top of the emotional burden, worry, and life & death uncertainty that patients and their caregivers fight through; it almost becomes criminal. Modern day data analytics and data tools have the potential to make the patients’ and caregivers’ life easier, and I would like to play my part in making it happen.
Q: Can you share an experience when you gathered data from multiple resources and combined it into actionable insights for your company. How did you determine which source was relevant and how good your product is performing?
I am happy to share some details of a data product called “Press Boost”, that I designed and implemented. But before going into the product detail, some background on the company, Doximity, is needed. A comparable sometimes used to introduce Doximity to an unfamiliar audience is like LinkedIn for doctors. Only verified physicians can join the network, making the conversations, discussions, and referrals safe and HIPAA laws compliant. Doximity aims to be the newsfeed of medicine, personalized for each verified physician. We have the largest readership in medicine, have an in-house editorial team, and analyze (and surface) 200K+ articles per week. “Press Boost” is a (free) data product that helps source articles with high medical value and engagement potential for our clients.
We ingest thousands of medically relevant articles daily from hundreds of online news publications (for non-journal organic articles) and PubMed (for research-based journal articles). After ingestion, the articles go through various rounds of NLP and regex matching to determine their medical relevance and extract hospital and medical facilities mentioned in those articles. Successful completion of these steps gives us a mapping of how medically relevant an article is and which hospital systems and facilities it mentions.
The product that I built, “Press Boost,” helps source articles with high engagement potential. First, it tracks internal trending articles based on engagement on Doximity (clicks, likes, comments). It then also looks at external trending research articles by ingesting the Atlmetric score (how much and what type of attention a research output has received) and the Mendeley readers score (how researchers engage with research on Mendeley). Finally, all of these factors are weighted and combined into scores. Such top-performing news content is then redistributed to clients associated with or interested in the hospital systems mentioned in those articles.
For interested readers, there is a beginner-friendly “Product Spotlight: Press Boost” presentation available on our website.
Q: Being a leading woman in health IT data science, how do you overcome career obstacles? Have you encountered gender discrimination or sexism?
I face the same challenge that every other woman in the technology world faces, the dichotomy of dual expectations. This is best explained by Dr. Deborah Gruenfeld, a social psychologist and professor at Stanford Business School. She defined the dual expectations as playing high, which means you show your authority, power, influence, and playing low, which means you are more approachable and likable. (See her video explaining the concept here). As leaders in the technology field, we are expected to play high, but as women, we are traditionally expected to play low. So, when we play high, we are deemed not likable, and when we play low, we are considered to be not competent! It’s a continuous balancing act. Patience, good judgment, and wisely picking my battles have been my friends in this journey.
As far as gender discrimination and sexism is concerned, there is no denying that it exists. I, however, have a certain approach to dealing with it. I read this great quote by Deepak Chopra “What you pay attention to grows. If your attention is attracted to negative situations and emotions, then they will grow in your awareness.” Hence, I don’t actively scan for discriminatory behavior, as that will put you too much into the fight or flight zone (thanks amygdala!); however, if discrimination openly finds me, I fight against it with all my strength, courage, and prudence.
Q: Being associated with ACM (as a senior editor) and ACM Women (as the standing committee’s chair), what are some of the contributions you are most proud of?
I am a keen enthusiast of promoting diversity in the STEM fields, especially encouraging women to be a part of the evolving disciplines of Computer Science and Data. I am a volunteer at Association for Computing Machinery – Women, Computing Research Association- Widening Participation, Rewriting the Code, TechGirlz, MentorNet, to name a few, and some regional groups like Pakistani Women in Computing and WomenInTechPK. Some of the most rewarding experiences in my life have been as a volunteer.
I am a senior editor for Ubiquity, ACM’s peer-reviewed web-based magazine devoted to the future of computing and the people who are creating it. I have been the first female member of our editorial board and started a new section, “Ubiquity: Innovation Leaders,” which consists of interviews with young professionals who comment on their concerns about the future of computing and their ambitions to shape the future through their leadership. As a result, I have been able to present several moving and compelling stories from diverse backgrounds and computing disciplines.
I am also the Standing Committee’s Chair for ACM-W. There I got the opportunity to propose a new initiative, a web series, “ACM-W: Celebrating Technology Leaders.” The idea is to bring stories and advice from engaging speakers, women with diverse careers in computing, directly to our global audience.
Q: Pakistan has a severe lack of peer-reviewed and general science magazines. What role do you think magazines like Scientia Pakistan can play in promoting science writing culture in Pakistan, especially among university students?
I genuinely believe that Scientia, with its mission to re-shape the narrative of science journalism in Pakistan, is doing an excellent service not only for the Pakistani student community but for a global audience.
Science is not an elitist club’s game, and good writing is not an outdated skill. Scientia is working towards mitigating both these misconceptions. Science is ubiquitous, everywhere working for everyone; hence it should be accessible to everyone. One’s writing reflects one’s personality, and “unedited incoherent streams of consciousness riddled with cyberslang, shorthand, and emojis” is not an attractive personality type. Scientia enables, encourages, and brings quality scientific writing to the masses without compromising either accessibility or scientific merit. Kudos to the entire team!
I believe active partnerships with leading universities and software houses in the country, encouraging academicians and practitioners to contribute regularly, will increase the visibility and impact of the initiative.
Q: How do you see the field of health data evolving?
Data Science, combined with machine learning & artificial intelligence advances, has enormous potential for improving the health industry. Data science can improve the speed and accuracy of testing and diagnosis, improve health research and drug development, strengthen diverse public health interventions, etc., and never has the utility been more apparent than in the COVID-19 era. e.g., in the last year and a half, data has been extensively used to
- Understand and predict the pandemic spread (using principles from network science, econometrics, applied microeconomics, etc.)
- Create effective treatments, creating algorithms capable of computationally generating, screening, and optimizing hundreds of millions of therapeutic antibodies (gene sequencing, computational biology, etc.)
- Resume and maintain economic activities (epidemiological modeling, epidemic dynamics, social networks, time series analysis, agent-based, network simulation, complex systems, etc.)
- Track spatial distribution of COVID-19 (contact tracking, geo-visualization, spatial data science, digital biomarking, remote sensing, etc.)
- Understand the evolution of hate speech, misinformation (large-scale measurements, social media, game theory, etc.)
However, the value proposed can only be realized if ethics, empathy, and civil liberties are at the core of the algorithmic design, data modeling, deployment and analytics usage. Two of the most significant issues in the data world are (1) unethical collection and use of patient data and (2) biased algorithms. World Health Organization (WHO) has recently released WHO: Ethics and Governance of Artificial Intelligence for Health Report (June 28th, 2021) that identifies six principles to ensure AI works to the public benefit of all countries. These principles are protecting human autonomy, providing informed consent, quality control, and transparency of the algorithms, inclusiveness irrespective of age, gender, ethnicity, etc., and transparent continued monitoring during actual use. I firmly believe these are indeed the six areas of future growth, research, and practice in the field of health data.
Q: Do you think that there are significant opportunities for women data scientists in healthcare management?
As I mentioned before, data science faces two significant challenges: unethical collection and usage of data and biased information. A major source of bias in many datasets is that the people who collect, organize and analyze the data do not represent the people that will actually be using the technology. For example, according to Harnham’s Diversity Report for the US Data & Analytics industry 2020-2021, women hold only 18% of the data science jobs, and the problem is likely worse in most lower-income countries. Data, in most cases, is like Rorschach charts; people see their own values, interests, and experiences reflected in them. If not careful, this opens the door to bias at every stage of the data value chain (sourced from Open Data Watch).
The underrepresentation of women, or any demographics, in data science increases the possibility that the data-driven decisions and products will not represent their interests or, in extreme cases, may harm their interests. I would highly recommend reading Carolina Criado Perez’s award-winning book “Invisible Women: Exposing Data Bias in a World Designed for Men” which talks about the issue in detail. One of the best ways to mitigate bias is to make sure that the data team consists of diverse experiences and perspectives to begin with. There is a global business realization that interpreting causal relationships and correlations in large data sets requires subtlety, and women bring different intuition to the table. The field of data science is exploding with opportunity. So yes, not only do women have a lot of scope in the field of data science, but this may be one of the best times to enter the field fueled by COVID-19.
Q: What advice do you have for future data scientists, especially women?
My advice is a little broad that any young man or woman in the STEM field may be able to gain benefit from, should they agree with my point of view of course.
We are at the brink of the fourth industrial revolution, powered by a fusion of technologies that are quickly blurring the lines between real and virtual, physical and digital. We need to guide and inspire a tech workforce ready for this unprecedented, disruptive future where quick obsolescence may be the biggest threat and remaining relevant, the biggest struggle. The most important training in this regard is to help future STEM professionals grow a generalist mindset. Rather than being tied to or specialized in a particular language, framework, or solution, generalists have a basic working knowledge of multiple domains, principles, and technologies. This helps them remain relevant in a variety of engineering jobs and projects. Moreover, they know “how to learn” and thus can quickly come up to speed and morph as per given technical preferences and constraints.
Second, this I share, especially for the women readership, you don’t have to negate parts of your personality to be perceived competent. I have shared this advice before, but I believe it cannot be reiterated enough times. For example, have you heard (admittedly well-meaning) statements like, “sure, humility is a good value, BUT it’s time to set it aside and work on self-branding”? It doesn’t have to be one or the other. You can be humble AND work on self-branding. You can be polite, flexible, and accommodating AND not let people take advantage of you. You can use “sorry” in your conversation and “just” in your emails if it is part of your politeness language, AND make sure you are being taken seriously. Whatever comes naturally to you, whatever feels authentic, is ok to hold on to while still evolving to a better self. If you want to change and leave some personality traits behind, that is fine, too, as long as you don’t feel obligated to do so.
Dr. Bushra can be reached out on Twitter @DrBushraAnjum or via her website https://www.bushraanjum.info/
Maham Maqsood is the Managing Editor at Scientia Pakistan. She has done her Bachelors from Quaid-i-Azam University in Biochemistry. An avid reader and a freelance writer, Maham has worked for several organizations including Globalizon and MIT Technology Review Pakistan.