In reality it is the intensity and/or duration of these somatic symptoms and not merely their presence that differentiates a person with CFS from a healthy person. Further, it is important to elicit self-report data using structured interview schedules. This ensures

that GSK1120212 cost questions are presented uniformly and avoids variable patient responses based on how questions are phrased. The CDC Symptom Inventory assesses information about the presence, frequency, and intensity of 19 fatigue related symptoms during the past month (Wagner et al., 2005). All eight of the critical Fukuda et al. symptoms are included as well as 11 other symptoms (e.g. diarrhea, fever, sleeping problems, nausea etc.). Jason et al.’s (2010) DePaul Symptom Questionnaire provides another structured Ponatinib molecular weight way to gather standardized information that can be used to aid diagnosis using the 2003 Canadian criteria (Carruthers et al., 2003) for what is termed ME/CFS. When categories lack reliability and accuracy, quality of treatment and clinical research can be significantly compromised. If CFS is to be reliably described by the clinical and scientific communities, it is imperative to deal with criterion

variance issues and provide specific thresholds and scoring rules for the selected symptomatic criteria. The same issues are relevant to other aspects such as characterizing CFS disability (Jason et al., 2011b, Reeves et al., 2005 and Wagner et al., 2005). In addition, instead of thresholds and a yes/no scoring of symptoms, the use of a continuous scale might address some of the issues that arise with conventional cohort stratification. Data mining, also referred to as GPX6 machine learning, might in the future help determine the types of symptoms that may be most useful in accurately describing CFS.

Data mining is a technique to explore large sets of data and either (1) replicate human decisions, especially when the process by which these decisions are made are not well-understood or (2) uncover patterns in the data that would not be evident to humans because of the size and complexity of the data. In the particular case of identifying CFS symptoms, both goals are desirable; using data mining to augment physicians’ diagnoses could result in more uniform diagnoses, while understanding symptoms most important in the diagnosis process could allow researchers to focus attention on the evaluation of those symptoms. Decision trees attempt to predict a classification for each patient based on successive binary choices: at each branch point of the tree, all the symptoms are examined with respect to their effect on the entropy of the diagnoses. Symptoms with high entropy are deemed important, and used to split all the cases into two parts.