Saturday, July 30, 2016

Practical LaTeX for the Health Sciences

Download full PDF article (first section printed below)
The purpose of this tutorial is to introduce health scientists, analysts, and writers to LaTeX for preparing scientific documents. LaTeX is a document preparation system for creating professionally typeset scientific documents.  LaTeX is freely available and widely used by data scientists, mathematicians, physicists, statisticians, engineers, demographers, and many other disciplines. Specifically, we will learn how to prepare a scientific article, report, and doctoral thesis. Additionally, we introduce selected software solutions that enhance the publication process.

The purpose of scientific writing is to communicate, persuade, educate, inform, or alert readers using content that is well-organized and clear.  Scientific and technical documents can be divided into the following components (in order of importance!):

  1. Content
  2. Structure
  3. Appearance

The document content is the main reason for writing anything: we want to effectively communicate, and perhaps persuade, our audience with our narrative and supporting tables and figures.  As writers, we want to spend our time and intellectual energy producing excellent content.  Next in importance is document structure: that is, how our document is organized for logic and flow: title, section headings, subheadings, bibliography, tables, figures, etc.  Good document structure optimizes the logic and flow of our content.  Last in importance is document appearance. We do not want to waste our time worrying about how the content will appear---this can be accomplished efficiently later if the document is well-structured to begin with.

Therefore, as we write, we should spend most of our time on content production, spend time on determining organization to optimize the order and flow of our content, and spend minimal time on formatting appearance.  All too often writers spend an extensive amount of time formatting the appearance of their document to give it a desired structure and appearance.  This is problematic for documents that are long or that require frequent updating.  Additionally, most writers are not trained in typography: the time wasted on formatting is much better spent on improving content.

Preparing scientific documents is not writing a fiction novel.  In many ways preparing a scientific document is easier.  First, the organization has an expected structure. For example, a scientific article generally has the following sections: introduction, methods, results, and discussion.  Second, scientific writing should be factual, concise, and clear.  And third, displays are generally limited to tables and figures.

Preparing scientific documents present the following challenges:

  • Organization is structured (introduction, methods, results,  etc.)
  • Document length may be long
  • Document may require periodic updating
  • Use of mathematical notation and equations
  • Management of references
  • Creation of bibliographies
  • Cross-references to equations, tables, and figures
  • Re-number equations, tables, and figures
  • Generation of table of contents, tables, and figures 

Because article manuscripts are relatively short (about 20 pages double-spaced), these issues are less problematic.  For a doctoral thesis (or long report), these issues are either addressed efficiently and save time, addressed incorrectly and waste someone's time (possibly an administrative assistant---or worse, the author---spending hours reformatting), or not addressed at all---resulting in a lower quality, less user-friendly document.

In general, document preparation systems can be classified as either visual design or logical design. Microsoft (MS) Word is a familiar example of a visual design system; it is also known as "what you see is what you get" (WYSIWYG).  What you see on the computer screen is almost identical to what you get when you print the document.  In MS Word, basic word processing is easy to learn, user-friendly, and convenient for local formatting.  Local formatting is achieved by highlighting a string a text and then formatting it: for example, italicizing, bolding, or changing font face, size or color.  A major limitation of visual design is that local formatting of appearance is so easy that it becomes (unintentionally) the formatting method of choice for structuring long documents.  Extensive local formatting of long documents become onerous and impractical; WYSIWYG comes to mean "what you see is what you got."

In contrast, logical design separates the process of content production from the processes of formatting structure and formatting appearance. LaTeX is a logical design system: it provides a "markup language" to mark up content to have structural and conceptual meaning.  This facilitates global formatting of structure and appearance using established typographical standards for scientific documents.  Once the content is marked up, the content is compiled into a professionally typeset document using well established formats for scientific publication.  The writer spends little time worrying about formatting structure and appearance, and more time on preparing high quality narrative content.  The best way to understand this is to experience it firsthand.
Download full PDF article (first section above) 

Thursday, July 14, 2016

Saturday, June 11, 2016

Cultural Humility --- and why it matters!

Cultural Humility

My road to discovering cultural humility was long, complex, and iterative. I have struggled with how to use our technical expertise to address health inequities and reduce health disparities. In April 2014, the San Francisco Department Public Health launched the Black/African American Health Initiative (BAAHI) which required me to question my assumptions and redefine my role. This happened by improving how we listen to our African American staff and communities.

I initially focused on technical solutions, but quickly learned (again!) that technical solutions---no matter how great and well-intentioned---will not take root and spread if we do not address underlying mental models, and explicit and implicit biases. Through an iterative process BAAHI emerged into three components:
  • Cultural humility (focused on racial humility)
  • Workforce development
  • Collective impact for health disparities
Cultural humility brings together and synergizes two important concepts: culture and humility. Culture is a "set of patterns of human activity within a social group and the symbols that give such activity meaning. Customs, laws, dress, architectural style, social standards, religious beliefs, and traditions are all examples of cultural elements. At every level, societies to individuals, culture is multi-dimensional and each individual has their own unique, multi-dimensional expression of culture which is dynamic and changing. Much of culture is hidden: we only see the symbols (behaviors, words, customs, traditions) but not the underlying beliefs, values, assumptions, and thought processes.

Humility is "the noble choice to forgo your status, and to use your influence for the good of others before yourself" (John Dickson). Humility is a universal character virtue because it is positively valued in nearly every society, culture and religion, and throughout modern history. Cultivating humility enables one to seek honest, critical feedback, and to improve relationships, trust building, team performance, and intellectual growth.

In 1998, Melanie Tervalon and Jann Murray-García published a groundbreaking article that challenged the concept of "cultural competency" with the concept of "cultural humility." When you accept cultural humility, by definition, you acknowledge that you can never truly achieve cultural competency. Cultural humility is committing to lifelong learning, critical self-reflection, and continuous personal transformation.

Here is my synthesis of their classic paper on this concept:
  1. Commit to lifelong learning and critical self-reflection
  2. Realize our own power, privilege, and prejudices
  3. Cultivate humility and empathy for respectful partnerships
  4. Recognize and validate our common humanity 
  5. Practice humble inquiry and deep listening
  6. Promote institutional accountability
I have come to believe that cultivating humility and practicing humble inquiry is central to leader, team, and organizational learning, performance improvement, and transformation. Organization culture is transformed through relationships (dyads and teams). Prejudices include explicit and implicit biases.

Humility and humble inquiry builds trust. Trust enables cooperation. Cooperation is necessary for shared visioning, shared decision making, and shared learning.

I believe that promoting cultural humility is one path to organizational transformation that will enable continuous improvement internally and externally with the diverse communities we serve.

For a more information on cultural humility visit:

Saturday, May 21, 2016

Population health data science, complexity, and health equity---Reflections from a local health official

I was honored to be invited to speak at the Stanford Center for Population Health Sciences Annual Colloquium on October 26, 2015. They have an exciting new transdisciplinary program to improve the health of populations.

I covered three related areas: population health data science, complex adaptive social systems, and health equity. I ended my talk by focusing on the inter-generational, lifecourse transmission of the effects of trauma to children ages 0 to 5 and how this contributes to racial health inequities.

Here was my thesis:
  1. For the new field of population health data science we must focus on transforming health relevant data into actionable knowledge. To produce actionable knowledge we must integrate methods from the fields of human-centered design, decision sciences, and behavioral economics. In a sense we must start backwards. In the traditional approach we focus on studying populations to discover average solutions ("one size fits most"). Actionable knowledge must be user- and context-centered, and account for individual variation. We use emerging technologies to make this faster, cheaper, better, and actionable.
  2. Human populations are complex adaptive social systems (CASS). To tackle the toughest challenges (e.g., health inequities) requires CASS approaches. How we conceptualize CASS frame how we study and test solutions. To understand CASS we must leverage computational modeling. From this laboratory we learn that simple rules can produce very complex phenomena. We are humbled to know that existing “real world” data are only one realization of many possible realizations. "Off the shelf" solutions do not exist; we must iterate to a solution through community engagement, experimentation, and continuous improvement (e.g., collective impact). 
  3. Our ultimate goal is to mobilize and transform communities. This starts by learning how to transform ourselves, our teams, and our organizations through testing and learning (“continuous improvement”). In public health the approaches we use to transform communities and inform policy decisions include health impact assessment, collective impact, and community-based participatory approaches. These key approaches add capability to our toolbox public health methods. Collective impact for transforming CASS problems has received enormous attention---and deservedly so. Collective impact is continuous improvement methods applied at a social scale for CASS problems. Collective impact embraces the challenges of complexity, and methodically focuses on collaboration and community transformation through a common agenda, shared measurement, mutually-reinforcing activities, continuous communication and improvement, and backbone support.
  4. Our main population health agenda must be to eliminate racial health inequities. Our collective priority must be to interrupt the inter-generational transmission of the effects of trauma (toxic stress) on young children. Toxic stress alters the brains, bodies, and behaviors of young children, thereby permanently affecting memory, judgment, self-regulation, and physiology. This results in higher risk behaviors and adult chronic diseases. Furthermore, Black/African Americans are traumatized throughout their lives by racism and discrimination. This focus---inter-generational transmission of toxic stress---enables us to prioritize and target social policies and social determinants of health with the biggest potential to eliminate the “childhood roots of adult health inequities.”

Summary slide

  1. Population health data science
    1. Start backwards (understand individual and group decision-making!)
    2. Focus on actionable knowledge (Advise--Predict--Discover--Describe)
    3. Focus on human-centered design (“precision public health”)
  2. Transforming complex social systems
    1. Understand complex adaptive systems (requires humility)
    2. Transform self, teams, organizations, communities (in that order:
    3. requires continuous improvement, taking risks, learning from failures)
  3. Tackling population health inequities
    1. Inter-generational transmission of trauma
    2. Toxic stress alters brain, body, and behavior
    3. Life course of trauma, racism, and discrimination
    4. 4Ps of public health: prevent, protect, prepare, promote
    5. 6Ps of complex systems: people, policy, place, program, provider, parents

Slide presentation

Video presentation

Saturday, May 14, 2016

Saturday, May 7, 2016

What Stephen Curry can teach us about the science of improvement

The Golden State Warriors can teach us a few things (actually a lot!) about leadership, organizational culture, teamwork, and performance improvement (also called continuous quality improvement). Performance improvement consists of improving processes to deliver better results. Improvement can be incremental or breakthrough. The science of improvement is best understood by understanding the theory of knowledge creation (plan-do-study-act) and single-loop and double-loop learning (Maccoby 2013).

On April 16, 2016, the New York Times published a graphic (Figure 1) depicting "752 lines---one for each NBA player who finished in the top 20 in 3-point attempts made in each season since 1980. Sitting atop it is the Golden State Warriors's Stephen Curry, who finished the regular season with a record 402 3-pointers" (Aisch, 2016).

Figure 1: Stephan Curry's 3-point record in context is "off the charts." "This chart contains 752 lines --- one for each NBA player who finished in the top 20 in 3-point attempts made in each season since 1980. Sitting atop it is the Golden State Warriors's Stephen Curry, who finished the regular season with a record 402 3-pointers." (Source: [])

Incremental performance improvement occurs by improving practices, and practices are based on accepted theories. A theory is an explanatory (or causal) model that can explain observed phenomena. Theories are not always explicit; they can be assumptions or mental models, sometimes they are hidden. The typical approach is to use PDSA cycles to test and adjust practice improvements (Figure 2). We plan to test a practice innovation, we test (do) the practice innovation, we study the results, and we act on what we learned, leading to incremental improvements.

Figure 2: Plan-Do-Study-Act (PDSA) cycle of experimentation and continuous improvement
Chris Argyris called this single-loop learning (Maccoby, 2013). He recognized that PDSA can also be used for double-loop learning which can lead to new theories and breakthrough performance improvements. Figure 3 depicts PDSA with single-loop and double-loop learning.

Figure 3: PDSA with singe-loop and double-loop learning

For example, when we are dissatisfied with the results of a practice we have two choices:
  1. Improve the practice (single-loop learning; possible incremental improvements), or 
  2. Improve the theory (double-loop learning; possible breakthrough improvements)
Double-loop learning makes these possibilities explicit and encourages innovative (breakthrough) thinking.

I propose that before Stephen Curry, we witnessed primarily single-loop learning and incremental improvements in making 3-pointers. Note that the cumulative improvements over three decades is very impressive! 

Also, I propose that with Stephen Curry (supported by an amazing team and organization), we are witnessing primarily double-loop learning and breakthrough improvements in making 3-pointers. Curry is changing the physics of shooting, and the Warriors are changing how basketball is played. The Warriors are developing new theories of shooting and playing. Many articles have been written about the Warriors's record setting year in basketball.

I often use sports to illustrate continuous improvement. I now use Figure 1 to teach not just PDSA, but also single-loop and double-loop learning. After all, single-loop and double-loop learning makes PDSA much more powerful, practical, and fun. Double-loop learning enables us to always question our assumptions, and is most productive when we use diverse, transdisciplinary teams.


Aisch G and Quealy K. Stephen Curry's 3-point Record in Context: Off the Charts. New York Times. April 16, 2016. Available from:

Maccoby M, Norman CL, Norman CJ, Margolies R. Transforming Health Care Leadership. 1st ed. Jossey-Bass; 2013. Available from: