Saturday, December 3, 2011

Increased arterial tortuosity in hypertensives

My co-authors and I published a peer reviewed journal paper developing a method to measure  arterial tortuosity and validated the method by measuring an increase in tortuosity in a hypertensive population. The idea is that vascular disease may twist the arteries and we developed a method to measure this twisting.

Other significant findings included the use of interpolating lower resolution images to higher resolution in order to compare images taking at different resolution. This is important because it allows comparison of existing images. Researchers won't have to reimage patients all at the same resolution reducing the cost of conducting research and clinically patients can be tracked over time even if different medical imaging equipment and resolutions are used. The full article is here:

Sunday, June 19, 2011

Liberal Sciences

The recession set off discussions on the value of a humanities or liberal arts education[1]. The argument against a liberal arts education is that students accumulate debt to get the degree and then can't get a job to pay it off. The argument for a liberal arts education is that it creates global citizens that can use their broad skills to do anything.

I don't think a liberal arts education is so broad compared to more technical or scientific degrees. Our world is very technological. How can an education be broad and enable students to go do anything if it doesn't include technical and mathematical skills?

Computers and the Internet are such a big part of our lives now. I think a liberal arts education should include learning the basics of computer programming because anyone in any career will use computer tools that have computer control interface that is a lot like computer programming. I have seen professors of music using computer software to compose music. English literature professors use natural language processing software  to identify plagiarism. The experts in the fields should be adept at using and modifying these tools because they know more about what the tools should be doing than a hired computer expert. The now renamed Racket programming language was created to make computer programming part of a liberal arts education and is the topic of the book "How to design programs."

Liberal arts prides itself on teaching analysis. Mathematics, statistics and data visualization are a great part of data analysis. Look at a field such as Economics. Economics is often called the dismal science. I think it is called this because the economics is so complex and includes a good dose of human psychology making it difficult for Economists to concretely test hypothesis and accept or reject them. Economics teaches strong quantitative analysis and data visualization skills because economists have to use everything to argue their points because the hypothesis often can't be conclusively accepted or rejected.

If a Liberal Arts education were converted to a Liberal Sciences education with courses in computer programming, statistics, data visualization, and scholars learned to make their arguments with statistics and visualizations of data graduates could go on to do anything in our increasingly technological world and the graduates would be very employable.

Alternatives to academic research for PhD

There are so many news articles lately on the value of college degrees. Of particular importance to me is the value of a science PhD. The basic problem is that there are too many science PhDs for traditional tenure track academic research positions. The number of PhDs generated at universities is based on available funding and the need for (cheap) labor for research projects, not on the job market for tenure track academic research positions. If PhD programs prepare students for other kinds of careers I don't think this is a problem. It really isn't a problem if PhD researchers are trained to transfer new technologies into new business ventures. That way some new PhDs will create new businesses to employ PhD scientists outside of academia.

I have seen some suggestions that universities should grant more masters and less PhDs. Then the masters can pursue research careers. The problem I see with this is the PhDs are more likely to gain the top leadership positions. This  would determine who gets the top posts early in the scientists careers and could hold back masters degree holders that are late bloomers.

I think a better solution is for departments to have tracks for PhD candidates to learn non academic research skills like technology transfer, business development, and management.

Saturday, April 9, 2011

Biologics are the future of drug development

Drug development expenses have been going up while FDA approvals of new drug treatments have been going down. We know more about biology now than in the past so this increased understanding should make us more efficient and effective at finding new and improved treatments. This could be a low hanging fruit problem where we found the easier treatments and new treatment are getting harder to find. One of the main problems stopping new drug treatment approvals is side effects.

To develop new drugs we identify a gene->protein target in a disease pathway, develop an assay that measures the activity of the gene->protein and test chemicals that down regulate or up regulate the gene target that affect other genes->proteins as little as possible. Finding a chemical compound that only affects the target and nothing else is quite difficult.

A non-repetitive sequence of 20-25 nucleic acid base pairs is unique in the human genome. This means a biologic drug like an siRNA, anti-sense RNA or gene therapeutic is our best method to uniquely target a single gene and best chance to reduce side effects of new treatments.

Medical informatics aligned with health costs

I have been asked if medical informatics, computers and software in medicine, the field I am getting a PhD in, can reduce medical costs. I think medical informatics CAN reduce medical costs with emphasis on CAN if and only if there is motivation to use computers and software to reduce medical costs.

Computer systems can track, remind, enable reuse of medical data like images, increase efficiency, reduce errors, determine the most cost effective treatments - reducing costs. All these good things ONLY end up reducing costs if our medical system is structured to motivate the stakeholders involved to reduce costs. I have heard one use of electronic medical records to increase the bill. It turns out when you go to the hospital and a dizzying assortment of physicians, nurses and medical specialists visit you and you receive a separate bill from almost every single person that saw you, they often forget to bill you for something. Electronic medical record systems can be used to make sure nothing was left off the bill and increase the cost to the patient.

What happened to efficiencies reducing medical costs? If a business increases efficiency but if the customers don't actively look for cheaper costs the business will simply keep the extra profit instead of passing the saving on to customer who is the patient in this case.

In what passes for a health system in the USA employers are given tax advantages for paying for employees health benefits instead of paying the employee the extra money and letting the employee buy their own insurance. The insurance is getting paid out of what would have been the employees' salary reducing the salary but the employees frequently don't know how much they are losing out on and how expensive the insurance actually is. Many higher level plans, especially the so call "Cadillac" plans only require a small co-pay for the patient. The insured patients don't care what health care costs because they have already paid for it through reduced salary. We can't expect the customer/patients who have already paid a very high price for health care to try to get a better deal on health care when it won't even reduce the patients' out of pocket expenses. The clinicians are busy treating patients and don't deal with bills much so they don't care about reducing costs either. The medical administrators that handle the business of health care don't care about reducing costs because the patients don't choose heath care providers on cost; clinics are not competing with each other on cost or efficiency. The insurance companies are about the only stakeholder interested in reducing costs and they can increase premiums to offset the higher medical costs to keep the same profit margin.

Employers should be taken out of the health care business if they are not actually health care businesses. It is not their business and I am sure they would like to focus on their business instead of being a health care insurer. They would have the added benefit of not having employees pick employers based on health care benefits which I know does happen. To get employers out of health care the tax deduction for employer based insurance should be ended. Hopefully the current health bill will allow private customers to purchase reasonable insurance. Co-pay should be a percentage of the bill or even the full cost up to some deductible limit, even a 5% co-pay on the full bill of  many treatments could bankrupt patients. This way patients know the full cost of their insurance and treatment and have to pay something based on the price so they are motivated on cost. Health care providers need to provide the costs to patients so that patients can compare providers easily. Emergency care is a different case, no one has time to shop around for emergency medicine. These costs would have to be controlled by negotiations between insurance and hospitals. After the people involved are motivated to control costs then medical information system case be used to increase effectiveness and reduce costs.

Medical costs can be reduced with the right motivation. Most types of medical cost are going up. Dentistry and especially plastic surgery are two types of medicine that are actually going down in cost while up in quality. They are also often not covered by insurance, patients pay most of the costs out of their pocket. Patients are motivated to find the best deal and providers are motivated to provide the best deal. Dentists and plastic surgeons are also known to use medical information system to make their practices run more efficiently, effectively and reduce costs and pass those reductions in cost on to the patients.

Sunday, April 3, 2011

Prefer building software libraries to platforms

Build software as libraries. Prefer libraries to platforms. A library is made up of reusable functions, classes and objects that software developers include in their projects. And multiple libraries can be reused in the software developer's projects. A platform is often designed as software that gives software developers lots of features to reuse but makes  software developers add their project to the platform; the reverse of the library.

Adding libraries to software developers' projects has many advantages to adding the project to the platform. Many libraries with different functionality can be combined together in the project. Adding a project to a platform means only the functionality of the platform can be reused. Platforms have another limitation. Functionality from a platform for building one kind of application, like command line applications is difficult to reuse in another kind of application, like a Graphical User Interface (GUI) or a web app or a cell phone app. Libraries let the software developers develop their application and reuse the the functionality in the libraries in their application.

When a platform is needed, build the platform using libraries that way the library can be used with the platform or without it. The ImageJ image processing program and library did this. The ImageJ libraries can be used in independent software applications, or the developer can build plugins (then ImageJ is acting as a platform) to ImageJ.

Sunday, March 13, 2011

Academy and industry growing the science pie

Like anyone who has worked in the commercial biotechnology science industry I have also spent time in academic science research in graduate school. There is definitely some rift between the two worlds of science. I find industry often doesn't know what is going on in nearby academic science centers, and academic science centers often don't want them to know because they think industry will steal their ideas just when they start working (There could be some truth to this.). The two science worlds should work closer together to grow the science pie.

The government invests in academic research to improve health, improve the environment, improve lives and stimulate the economy and stimulating the economy is getting top billing now. Government investment is a fixed amount allocated each year. Academic scientists write grants to compete for their share of the pie. When one scientist wins another loses. Industry starts with basic science and applies it to problems (like diseases) to make products and sell them increasing the amount of money coming into science, growing the pie.

Academic scientists can work on basic science questions with a long horizon before commercialization. Industry has to work on questions with a shorter time to market. The two will work on slightly different complimentary projects.

Industry can benefit from academic research opening up new areas for product development. Industry can provide academic researchers more places to pursue a science career when we finish graduate school. If every graduate stayed in academic research, grant competition would get even more fierce than it already is (and it is fierce). Industry could fund projects as they get closer to having commercial potential. I think companies will benefit most in the long run by not burning their bridges with academic research centers. Industry can provide funding (or licensing fees) to projects of interest and by being involved in academic research will know what new basic science is coming and start thinking about how to commercialize it. With the current and most likely increasing cumulative government budget deficit scientists may not always be able to rely on even the current level of government funding making it important for all scientists to get academy and industry to work together to grow their own science pie.

Monday, March 7, 2011

Medical imaging genomic disease subtypes

The genomics revolution is happening introducing personalized medicine but slower than most of us involved had hoped. One issue is that a disease have multiple genetic causes making traditional methods difficult. Medical imaging has the potential to separate subtypes of a disease based on what is seen in the medical image. Hopefully there are fewer genes involved in the medical image subtype making the genetic causes easier to identify. This explains why I have moved from genomic research to medical image processing. And I love creating these images.

Sunday, March 6, 2011

Contribute back; even for your own sanity

Open source software allows you to modify the code for your own uses. I use so many different open source tools that I haven't become a regular contributor to any one project. The regular contributors are amazing, I don't know where they find the time.

When you fix, modify or improve an open source software project, even if you don't become a regular contributor, send your changes to the project manager. People have been very good at including changes I have sent in.

This will help you in the long run. If you change your code and don't send it back the project will release a new version that doesn't have your changes. If you upgrade to the new version you won't have your changes and will have to make the change over again. Eventually  you will end up with a forked older version and cannot benefit from the improvements in the new versions. For an organization this can turn into a really large problem when they hire new computer programmers that are used to the newer versions of the software. When one piece of software depends on another you may end up not being to upgrade anything or start using new software that uses a newer version of the software you forked.

So send your changes in, even if you only a change an open source project once.

Saturday, March 5, 2011

Glass Research Towers instead of Research Parks

Research should be done in Towers not Parks to increase collaborations.

Let's call them Glass Research Towers instead of Ivory Towers which have such an uncollaborative reputation.

Buildings for research and technology venture companies should foster collaboration, Glass Towers can do this, Research Parks cannot. The typical Research Park is a large piece of land outside of town and one relatively low building is made for each institute or company. Companies or Institutes sign long term leases for the Park to make them a building. If an organization wants to expand it involves another long term lease and a long time to make a new low lying building. A small organization just getting started doesn't have the resources to sign a long term lease for an entire building. Small companies end up somewhere else in some kind of incubator.

The Park employees drive their cars into their Institute, work, eat in the company or drive out of park (through a few limited access road so there is lots of traffic) to lunch, work again and then drive home. They will probably never meet anyone from another building, not any fostering collaborations.After hours Parks are almost empty of people so few businesses like restaurants, that would provide a place for people for different Institutes to meet set up because their is so little dinner business.

The Parks could be a lot smaller in land area with a few tall Glass Towers. There need to be a few Glass Towers since some Research Organizations have heavy equipment like MRIs that would be hard to put in the higher floors. Smaller land requirements means the Glass Towers could be closer to town. The taller buildings would reduce the rental cost per foot ( For a description of tall building advantages see Organizations could still limit access to the their floors with the key swipe cards. Small unstable Organizations could get shorter leases. Growing organizations could lease new floors as needed (and shrinking Organizations could shrink as well and not be stuck a a large empty building they don't need).

Support organizations could provide services to multiple organizations in the Glass Towers when needed.  There could be software development companies, IT hardware support, computer help-desk, cloud computing, human resources, intellectual property and possibly venture capitol on site for the Research Organizations to hire when needed and not have to keep on staff when not needed. This could be a tremendous savings and convenience for smaller Organizations that cannot afford to keep all this in house.

There could be other non research support like restaurants, cafeterias, fitness clubs and day-care. Researchers would meet in these common areas fostering collaboration.  All of the support services would give small organizations most of the luxuries and strengths now only available to the largest organizations while maintaining small organization strengths. If there is any space left over put in a park instead of the the massive amount of parking lots needed by people driving between all of the spread out buildings.

Internet leads to in person collaborations

Collaborations have led to my best research results. Since my research involves developing software to analyze data I am always in need of useful data. I have received very interesting data from multiple sources. My collaborators and I will have publications out soon.

The Internet, web-sites and e-mails are useful for finding and conducting collaborations. Interestingly the best collaborations always get going after we meet in person. The e-connections actually help us meet in  person other researchers we normally never would have met. Being able to transfer and share data through the Internet makes continuing work possible but it also means meeting more researchers in person.