Scientists Simplifying Science

Tag archive

literature

How to create and measure innovation?

in That Makes Sense by

“Innovation” is THE buzz-word of today!

Everyone wants to label their companies as innovative, hire innovative people, create processes to induce innovation and be the next big innovator! But how does one really ‘innovate’ and how do we quantify innovation?

Sarah Kaplan, professor at the Rotman School of Management at the University of Toronto came up with an interesting answer that you might not have expected.

It is generally thought that brain storming with people from diverse knowledge backgrounds is a great way to come up with new ideas. In their paper, Kaplan and colleagues show that while combining different disciplines does lead to novel ideas, there is another equally important way that innovation works. In-depth knowledge in a field is required to understand the anomalies within the field, which can then lead to novel ideas.

“We find that, counter to theories of recombination, patents that originate new topics are more likely to be associated with local search, while economic value is the product of broader recombinations as well as novelty.”

Interestingly, breakthrough innovations were more likely to result from searches within a domain but economic value was a result of novel innovations arising from a combination of diverse ideas. However, such patents were very rare making up only 1% of the dataset.

“Patents that were both novel and had economic value were the most valuable. And that was only about 1% of the total patents.”

At this point, most researchers must be nodding in agreement “I had thought so”. What was the most surprising thing for me, though, was the way they measured “novelty”. In scientific literature as well as the patent world, innovation is measured as a direct function of citations. Even though most of the scientific community has rejected the idea of the journal impact factor as a way to measure the quality of a scientific article, the next best measure employed is the number of citations for the article itself. Thus, a patent or scientific paper that get highly cited is considered superior and thus a breakthrough innovation.

“What we found in our study is, in fact, that most of the patents that do get highly cited are not necessarily novel.”

In this study, the authors used a different metric to examine patents from the field of nanotechnology. A computer science and natural language processing (NLP) method called topic modeling that uses “a bag of words, a body of text, …and it infers from that body of text by the co-location of all the different words, what are the key underlying topics in the data”  was employed to determine if novel ideas were being developed. Interestingly, the patents that had high level of citations were not necessarily novel.

This is an interesting revelation, and something that scientists should also consider while judging the quality of literature. The entire reward system in science is largely based on publications and the feedback from citations. This generates ‘hot’ topics that many scientists work on, read about and cite, thus creating a research bubble. In such an environment, other fields of potential interest have difficulty to gain exposure and citations. Researchers flock towards hot topics, which can hinder the overall progress of science.

This generates ‘hot’ topics that many scientists work on, read about and cite, thus creating a research bubble. In such an environment, other fields of potential interest have difficulty to gain exposure and citations.

Kaplan and her group plan to delve deeper into how innovation works by studying novel ideas in different fields. It would be interesting to see what insights they can bring!

Me_DFX9112-crop

About the author: Czuee has a PhD from the University of Lausanne, Switzerland and Masters from IIT Bombay. She has previously worked at IISc-Monsanto collaboration and as a patent analyst at Evalueserve. Apart from her research on proteins involved in brain signalling and diabetes, she is interested in scientific communication (czuee.wordpress.com), entrepreneurship and runs a webcomic (http://gradschoolmuse-icals.thecomicseries.com/).

Photo source: Forbes.com

Creative Commons License
This work by ClubSciWri is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

Reading scientific literature – for dummies

in That Makes Sense by

If you are interested in the latest science breakthroughs, there is some good news for you. Be it to know more about a disease your mother has, to write a newspaper article about the latest scientific discovery, or to look for the latest research on a topic of undergraduate study; scientific literature is increasingly becoming available at your fingertips!
Recently, there has been a significant push in support of open access publishing in the scientific field. This “unrestricted online access to peer-reviewed scientific research”, as Wikipedia defines it, is supposed to create a revolution in scientific communication. The main argument in its favour is that research done with tax-payer’s money should be available to the public for free, and the internet has greatly made open access possible.
However, an important question that this movement has ignored, is this – how would the general public be able to understand, make sense of, and utilize this information in the right way? Scientific literature is often detailed and boring, with scientific jargon scattered all over, that only scientists from a specific field are able to understand. It’s a running joke between researchers that reading a scientific paper is the best cure for insomnia! So, how would a lay person be able to understand this information even if its made available to him?
If it helps, a paper on amygdala anatomy is as difficult to understand for me as it is for you, even though I am at the end of my PhD in life sciences – simply because I am in a completely different field. However, with some basic knowledge of science, you can actually manage to understand the scientific discoveries in a particular field if you have some time and patience.
In this post, I have given some pointers to simplify navigating through the ocean of scientific literature with very little or no knowledge of the field.

.
Don’t let the jargon scare you!

Sample this –

nrc3545-f4
Scientists are notorious for using the most complex-sounding tongue-twisters in their papers. But don’t let it deter you. It would help to have a medical dictionary or Wikipedia handy, to look up terms you don’t understand. The best papers are the ones where each term is introduced and explained properly. Also, try to understand that commonly used words have a slightly different meaning in scientific literature.

Where to look for an article?
For biomedical articles you can search on Pubmed, the repository of the NIH. A search for “alzheimer’s” gives all articles with the word anywhere in the paper. For more specific searches, you can use “alzheimer’s [Title/Abstract]”. You can also use other parameters for searching, like Author and date of publication and Boolean operators like AND, OR, NOT. Google scholar also gives good results.

If the article you are searching for is not available for free, try your nearest University library. Most Universities have access to paid journals. There are some online torrents, sites and even a Facebook page where academicians share papers that they have access to.

Start with reviews
Reviews combine the results from several articles to give a good overview of the field. If you are new to the field or want information in layman terms, reviews are generally a good starting point. Even as a researcher, reviews are the best place to start before we dive into the huge pool of literature on a topic. The downside is that you might not learn about fresh research that got published in the last few months. If a review seems too difficult, start with a textbook entry. There are many free books available on NCBI Books.

How to read a journal article
A research paper has several sections, generally arranged in a particular order – Abstract, Introduction, Methods, Results, Discussion and Conclusion. However, if you want to find information quickly without delving into all the details, you can follow this order:

Read the Abstract first and the Conclusion – this will give you a quick idea about what the paper is all about, and the highlights of the research presented.

Then read the Introduction and Discussion – This will give you an idea of previous research in the field. The ‘Introduction’ introduces the topic, and cites related literature. The ‘discussion’ explains the implications of the research carried out by the authors, and how it relates to the currently available knowledge.

And finally the Results section, which details the experiments performed and the exact outcome of all experiments. If you must read this section, start with analyzing each figure and table, and read their captions. Try to draw your own conclusions, and mark the irregularities in the data. Be critical of what is represented in the figures and what is claimed in the text, and if they indeed match!

Keep in mind the basic rules of statistics when analyzing data:

  • What information is represented on the X and Y-axes on graphs? Then think about what the data really indicates.
  • What is the sample size used for the study? In most cases, the bigger the sample size, the more reliable the results.
  • Is there any apparent bias in the sample? For example, did they consider only Caucasians for the study. In that case, is it still relevant for Asians?
  • Are there multiple experiments pointing to the same conclusion?

The Methods section is best left to a trained researcher in the field, and is generally not required to be understood to know the implications of the research.

Understand the general consensus in the field
If you are trying to find out if product x causes cancer, try to read several articles on the topic. This will give you an idea of what the best researchers in the field think about the topic. This way, you can also learn about different perspectives – one researcher might claim it causes cancer, while another might show it causes cancer only under certain conditions. Also, make sure that the articles are not from the same laboratory. For Biomedical literature, you can check the last author(s), who is generally the Principal Investigator (or Professor), and their University affiliations. If the articles published by different authors at different universities have similar conclusions, it shows that the research is reproducible and there is a consensus.

Impact factor is not the best indicator of quality
This is a mistake that I made until I was a Masters student! The impact factor of a journal indicates how well-cited the articles from this journal are. However, the most highly rated journals are like newspapers looking for the most sensational news. Although, most research published here is of good quality, it is best not to take it as the absolute truth. Instead, try to look for how well-cited and well-researched the article itself is. When you read an article in Pubmed, you can see how many times this article has been cited by others.

Take it with a pinch of salt
Scientific research is complicated and difficult, because one is charting new territories. As my adviser says, researchers are like the blind men trying to figure out an elephant. Each has their own version of the “truth” and all of them could be correct, or wrong! Hence, be careful about jumping to strong conclusions. High impact, well cited articles could be proven wrong or insufficient at a later point.

First published on LinkedIn on Jan 27, 2015

 

Me_DFX9112-crop

About the author: Czuee has a PhD from University of Lausanne, Switzerland and Masters from IIT Bombay. She has previously worked at IISc-Monsanto collaboration and as a patent analyst at Evalueserve. Apart from her research on proteins involved in brain signalling and diabetes, she is interested in scientific communication, entrepreneurship and runs a webcomic (http://gradschoolmuse-icals.thecomicseries.com/).

Creative Commons License
This work by ClubSciWri is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

Go to Top