Qualitative Research Methods to Conduct Needs Assessment AND Quantitative Evaluation of Negative Abstract Sets in Identification of Potential Biomarkers for Lung Cancer

Seminar Date: 
2012-04-13
Seminar Time: 
11am - 12pm
Seminar Location: 
M-184 VALE, 200 Meyran Avenue
Presenter: 
Jhon Camacho Sanchez & Rick Jordan
Presenter's Institution: 
TBA

Title:  Qualitative Research Methods to Conduct Needs Assessment

Jhon Camacho Sanchez, Masters Fellow


Abstract: A need-assessment study is a necessary first step in the path to design a clinical informatics intervention. However, this type of study present particular challenges due to the inherently subjective and multifaceted data required to answer its research questions. Questions like: what information the clinicians need  to perform their work? What are their expectations about the introduction of computers in their clinical environment? Or Which are the major barriers to improve the quality of care? are not amenable to quantitative research techniques. In this presentation, I briefly review some of the qualitative research methods to explore informatics needs in a clinical environment.

 

Title: Quantitative Evaluation of Negative Abstract Sets in Identification of Potential Biomarkers for Lung Cancer

Rick Jordan, MS, Doctoral Fellow
 

Background

Proper predictive experimental design requires a minimal sample size, as well as training and test sets, with each training and test set usually containing positive and negative samples. This has been well documented, and is standard protocol in the experimental realm. In this work, we evaluate the effect of both positive (keyword “Lung Cancer”) to negative abstract (‘NOT Lung Cancer”) ratios, and a minimum range of negative abstracts required, on the potential significance score for a marker. We incorporate this knowledge into a method to identify potential biomarkers found in the literature, that co-occur in an abstract with a combination of ‘lung cancer’ and 14 different biofluids.

Results

PubMed was queried to obtain the pertinent abstracts. Biological entities such as proteins and genes within the abstracts were tagged using ABNER, and further processed using python scripts to produce a final list. Significance scores were calculated, and compared to a list of known cancer biomarkers. Our results show that a positive-to-negative abstract ratio in the range of 1:10 to 1:25 encompasses significant scores over the greatest number of biofluids tested.

Conclusions

A recommendation of a minimum range of 6,000 - 6,500 negative abstracts, that are required to obtain the most significant scores, is given. We have determined a list of potential biomarkers for lung cancer, and will continue our work obtaining similar results for other diseases. 

^