Laboratory testing is the single highest-volume medical activity, making it useful to ask how well one can anticipate whether a given test result will be high, low, or within the reference interval (“normal”). We analyzed 10 years of electronic health records—a total of 69.4 million blood tests—to see how well standard rule-mining techniques can anticipate test results based on patient age and gender, recent diagnoses, and recent laboratory test results. We evaluated rules according to their positive and negative predictive value (PPV and NPV) and area under the receiver-operator characteristic curve (ROC AUCs). Using a stringent cutoff of PPV and/or NPV≥0.95, standard techniques yield few rules for sendout tests but several for in-house tests, mostly for repeat laboratory tests that are part of the complete blood count and basic metabolic panel. Most rules were clinically and pathophysiologically plausible, and several seemed clinically useful for informing pre-test probability of a given result. But overall, rules were unlikely to be able to function as a general substitute for actually ordering a test. Improving laboratory utilization will likely require different input data and/or alternative methods.
References
[1]
Arnaout R (2011) Big Data in Clinical Pathology. Critical Values 4: 15–19.
[2]
Rang M (1972) The Ulysses syndrome. Can Med Assoc J 106: 122–123.
[3]
Jackson BR (2008) The dangers of false-positive and false-negative test results: false-positive results as a function of pretest probability. Clin Lab Med 28: 305–319, vii. doi: 10.1016/j.cll.2007.12.009
[4]
Zhi M, Toupal-Theisen J, Feigl-Ding EL, Whelan J, Arnaout R (2013) The Landscape of Inappropriate Laboratory Testing: A 15-Year Systematic Review and Meta-Analysis. PLoS One doi: 10.1371/journal.pone.0078962
[5]
Tugwell P, Dennis DT, Weinstein A, Wells G, Shea B, et al. (1997) Laboratory evaluation in the diagnosis of Lyme disease. Ann Intern Med 127: 1109–1123. doi: 10.7326/0003-4819-127-12-199712150-00011
[6]
Arnaout R (2012) Elementary, my dear Doctor Watson. Clin Chem 58: 986–988. doi: 10.1373/clinchem.2011.180992
[7]
Nelder JA, Wedderburn RW (1972) Generalized linear models. Journal of the Royal Statistical Society Series A (General) 370–384. doi: 10.2307/2344614
[8]
Clifford L, Singh A, Wilson GA, Toy P, Gajic O, et al. (2012) Electronic health record surveillance algorithms facilitate the detection of transfusion-related pulmonary complications. Transfusion doi: 10.1111/j.1537-2995.2012.03886.x
[9]
Kitsantas P, Hollander M, Li L (2006) Using classification trees to assess low birth weight outcomes. Artificial intelligence in medicine 38: 275–289. doi: 10.1016/j.artmed.2006.03.008
[10]
Agrawal R, Srikant R (1994) Fast algorithms for mining association rules. Proc. 20th Int. Conf. Very Large Data Bases (VLDB).
[11]
Agrawal R, Mannila H, Srikant R, Toivonen H, Verkamo AI, et al. (1996) Fast Discovery of Association Rules. Advances in knowledge discovery and data mining 12: 307–328. doi: 10.1023/a:1009748302351
[12]
Mullins IM, Siadaty MS, Lyman J, Scully K, Garrett CT, et al. (2006) Data mining and clinical data repositories: Insights from a 667,000 patient data set. Computers in Biology and Medicine 36: 1351–1377. doi: 10.1016/j.compbiomed.2005.08.003
[13]
Stilou S, Bamidis P, Maglaveras N, Pappas C (2001) Mining association rules from clinical databases: an intelligent diagnostic process in healthcare. Studies in Health Technology and Informatics 1399–1403.
[14]
Weiss JC, Natarajan S, Peissig PL, McCarty CA, Page D (2012) Machine Learning for Personalized Medicine: Predicting Primary Myocardial Infarction from Electronic Health Records. AI Magazine 33: 33–45.
[15]
Bellazzi R, Zupan B (2008) Predictive data mining in clinical medicine: current issues and guidelines. Int J Med Informatics 77: 81–97. doi: 10.1016/j.ijmedinf.2006.11.006
[16]
Gibbs P, Turnbull LW (2003) Textural analysis of contrast-enhanced MR images of the breast. Magnetic Resonance in Medicine 50: 92–98. doi: 10.1002/mrm.10496
[17]
Wang T-L, Jang T-N, Huang C-H, Kao S-J, Lin C-M, et al. (2004) Establishing a clinical decision rule of severe acute respiratory syndrome at the emergency department. Ann Emerg Med 43: 17–22. doi: 10.1016/j.annemergmed.2003.08.002
[18]
Eftekhar B, Mohammad K, Ardebili HE, Ghodsi M, Ketabchi E (2005) Comparison of artificial neural network and logistic regression models for prediction of mortality in head trauma based on initial clinical data. BMC Medical Informatics and Decision Making 5: 3.
[19]
Kennedy EH, Wiitala WL, Hayward RA, Sussman JB (2013) Improved cardiovascular risk prediction using nonparametric regression and electronic health record data. Medical Care 51: 251–258. doi: 10.1097/mlr.0b013e31827da594
[20]
Kurt I, Ture M, Kurum AT (2008) Comparing performances of logistic regression, classification and regression tree, and neural networks for predicting coronary artery disease. Expert Syst Appl 34: 366–374. doi: 10.1016/j.eswa.2006.09.004
[21]
Westra BL, Savik K, Oancea C, Choromanski L, Holmes JH, et al. (2011) Predicting improvement in urinary and bowel incontinence for home health patients using electronic health record data. J Wound Ostomy Continence Nurs 38: 77. doi: 10.1097/won.0b013e318202e4a6
[22]
Harper PR (2005) A review and comparison of classification algorithms for medical decision making. Health Policy 71: 315–331. doi: 10.1016/j.healthpol.2004.05.002
[23]
Lehmann C, Koenig T, Jelic V, Prichep L, John RE, et al. (2007) Application and comparison of classification algorithms for recognition of Alzheimer's disease in electrical brain activity (EEG). Journal of neuroscience methods 161: 342–350. doi: 10.1016/j.jneumeth.2006.10.023
[24]
Breiman L (2001) Random forests. Machine Learning 45: 5–32. doi: 10.1023/a:1010933404324
[25]
Tapper EB, Rahni DO, Arnaout R, Lai M (2013) The overuse of serum ceruloplasmin measurement. Am J Med 126: e921–925. doi: 10.1016/j.amjmed.2013.01.039
