ADVERTISEMENT

Issue Navigator

Volume 12 No. 02
Earn CME
Accepted Papers
Classifieds





Commentary

Big-Data or Slim-Data: Predictive Analytics Will Rule with World

Daniel Combs, MD1,2; Safal Shetty, MD2,3; Sairam Parthasarathy, MD2,3
1Department of Pediatrics, University of Arizona, Tucson, AZ; 2Center for Sleep Disorders, University of Arizona, Tucson, AZ; 3Department of Medicine, University of Arizona, Tucson, AZ

Predictive analytics encompasses a variety of statistical techniques from predictive modeling, machine learning, and data mining that analyze current and historical facts to make predictions about future, or otherwise unknown, events.1 In medicine, the convergence of meaningful use of electronic medical records, ICD-10 diagnostic coding, data warehouses, and integrated healthcare systems are bringing such predictive analytics to the bedside and clinics in order to improve the health of the nation. The U.S. is investing significant amount of resources into the informational technology infrastructure with the intent of harnessing such big data to help predict, diagnose, and treat medical conditions and thereby improve population health. We need to strategically bring such resources to sleep medicine. In this issue of the Journal, Ustun and colleagues set us in such a deliberate direction by applying a new machine learning method known as SLIM (Supersparse Linear Integer Models). They tested the hypothesis that a diagnostic screening tool based on routinely available medical information would be superior to one based solely on patient-reported sleep-related symptoms. Their rationale was that the application of such technology—in an automated manner—to data residing in electronic medical records can assist with large-scale screening for obstructive sleep apnea (OSA).

In a recent review of epidemiological studies published over the last two decades, the prevalence of OSA, defined by polysomnography with an apnea hypopnea index greater or equal to 5/hour was estimated to be 22% in men and 17% in women.2 OSA is common, and the prevalence is rising3; however, OSA remains underdiagnosed.4 OSA is associated with a variety of negative health consequences, ranging from increased risk of hypertension, diabetes, cancer, and increased mortality.57 Accordingly, the American Academy of Sleep Medicine task force identified improvement in disease detection and categorization as one of many key outcome measures.8 Given these adverse effects, as well as the existence of effective treatments for OSA,9 there is a clear need for appropriate screening methods of patients at risk for OSA. In their article, Ustun et al. describe the use of machine learning techniques to develop screening algorithms for OSA10 in a sleep laboratory-based population. They tested whether the characteristics of their algorithms were comparable or superior to existing screening tools. Specifically, they compared the utility of patient-reported symptoms of sleep apnea (snoring, witnessed apnea, etc.) versus non-sleep specific factors (“extractable features”) that are typically available through an electronic medical record (age, body mass index, gender, and chronic medical problems such as hypertension). The approach of using a mix of reported sleep symptoms as well as non-sleep specific factors has been previously used in other screening tools, such as the Berlin questionnaire11 and the STOP-Bang.12,13 The authors found that the use of extractable features to develop a screening tool was significantly better than the use of symptoms alone. Additionally, the combined model utilizing both extractable features and symptoms was not superior to the use of extractable features alone.

All of the extractable features were intentionally selected to be readily available from the medical record. Although simple screening tools such as the STOP-Bang are available for screening in the primary care setting, an implementation gap exists with regards to routine screening for OSA or any other sleep disorder in the primary care setting.14,15 The lack of recognition of OSA in primary care is likely due to multiple underlying factors, including limited reporting of OSA symptoms by patients, limited visit time, as well as lack of provider knowledge. Research has shown that approximately 30% to 40% of primary care physicians' patients are at high risk for OSA based on the Berlin Questionnaire,16,17 and 90% to 99% may report a sleep related symptom when surveyed.18 Despite this high rate of symptoms when queried, only one in five patients had discussed their sleep concerns with their physician.16 Others have reported that only 7% of primary care physicians asked unsolicited questions regarding sleep.19

Prior research has shown that the use of a chart-based simple reminder can improve screening for sleep disorders.20 Potentially, the tool described by Ustun et al. could be integrated into the electronic medical record, flagging high-risk patients and prompting physicians to further screen for OSA. These high-risk patients could then be referred for diagnostic testing for OSA. The approach to use a screening tool that is not dependent upon patient-reported sleep symptoms sidesteps the barriers for detection of OSA in the busy clinic setting. While management of OSA by sleep-certified physicians may confer an advantage over providers with no prior experience in managing patients with OSA, such automated electronic medical record based systems could assist with case-finding and conceivably be comparable between providers who are not experienced, nor received training, in managing patients with as yet undiagnosed OSA versus those managed by sleep-certified physicians.2123 Ustun and colleagues should be commended for bringing both big and slim data to our doorsteps. Future research is needed to determine the feasibility, cost-effectiveness, barriers to implementation, and patient-outcomes of integrating such predictive analytics into routine practice.

DISCLOSURE STATEMENT

Dr. Parthasarathy reports grants from NIH/NHLBI (HL095799), grants from Patient Centered Outcomes Research Institute (IHS-1306-2505), grants from US Department of Defense, grants from NIH (National Cancer Institute; R21CA184920), grants from US Department of Army, grants from Johrei Institute, personal fees from American Academy of Sleep Medicine, personal fees from American College of Chest Physicians, non-financial support from National Center for Sleep Disorders Research of the NIH (NHLBI), personal fees from UpToDate Inc., Philips-Respironics, Inc., and Vaopotherm, Inc.; grants from Younes Sleep Technologies, Ltd., Niveus Medical Inc., and Philips-Respironics, Inc. outside the submitted work. In addition, Dr. Parthasarathy has a patent UA 14-018 U.S.S.N. 61/884,654; PTAS 502570970 (Home breathing device) pending. The above-mentioned conflicts including the patent are unrelated to the topic of this paper. The other authors have indicated no financial conflicts of interest.

CITATION

Combs D, Shetty S, Parthasarathy S. Big-data or slim-data: predictive analytics will rule with world. J Clin Sleep Med 2016;12(2):157–158.

REFERENCES

1 

Encyclopedia WTF. Predictive analytics.

2 

Franklin KA, Lindberg E, authors. Obstructive sleep apnea is a common disorder in the population-a review on the epidemiology of sleep apnea. J Thorac Dis. 2015;7:1311–22. [PubMed Central][PubMed]

3 

Peppard PE, Young T, Barnet JH, Palta M, Hagen EW, Hla KM, authors. Increased prevalence of sleep-disordered breathing in adults. Am J Epidemiol. 2013;177:1006–14. [PubMed Central][PubMed]

4 

Costa LE, Uchoa CH, Harmon RR, Bortolotto LA, Lorenzi-Filho G, Drager LF, authors. Potential underdiagnosis of obstructive sleep apnoea in the cardiology outpatient setting. Heart. 2015;101:1288–92. [PubMed]

5 

Peppard PE, Young T, Palta M, Skatrud J, authors. Prospective study of the association between sleep-disordered breathing and hypertension. N Engl J Med. 2000;342:1378–84. [PubMed]

6 

Reichmuth KJ, Austin D, Skatrud JB, Young T, authors. Association of sleep apnea and type II diabetes: a population-based study. Am J Respir Crit Care Med. 2005;172:1590–5. [PubMed Central][PubMed]

7 

Marshall NS, Wong KK, Cullen SR, Knuiman MW, Grunstein RR, authors. Sleep apnea and 20-year follow-up for all-cause mortality, stroke, and cancer incidence and mortality in the Busselton Health Study cohort. J Clin Sleep Med. 2014;10:355–62. [PubMed Central][PubMed]

8 

Aurora RN, Collop NA, Jacobowitz O, Thomas SM, Quan SF, Aronsky AJ, authors. Quality measures for the care of adult patients with obstructive sleep apnea. J Clin Sleep Med. 2015;11:357–83. [PubMed Central][PubMed]

9 

Giles TL, Lasserson TJ, Smith BH, White J, Wright J, Cates CJ, authors. Continuous positive airways pressure for obstructive sleep apnoea in adults. Cochrane Database System Rev. 2006;CD001106. [PubMed]

10 

Ustun B, Westover MB, Rudin C, Bianchi MT, authors. Clinical prediction models for sleep apnea: the importance of medical history over symptoms. J Clin Sleep Med. 2016;12:161–8.

11 

Netzer NC, Stoohs RA, Netzer CM, Clark K, Strohl KP, authors. Using the Berlin Questionnaire to identify patients at risk for the sleep apnea syndrome. Ann Intern Med. 1999;131:485–91. [PubMed]

12 

Chung F, Yegneswaran B, Liao P, et al., authors. STOP questionnaire: a tool to screen patients for obstructive sleep apnea. Anesthesiology. 2008;108:812–21. [PubMed]

13 

Combs D, Goodwin JL, Quan SF, Morgan WJ, Parthasarathy S, authors. Modified STOP-Bang tool for stratifying obstructive sleep apnea risk in adolescent children. PloS One. 2015;10:e0142242. [PubMed Central][PubMed]

14 

Miller JN, Berger AM, authors. Screening and assessment for obstructive sleep apnea in primary care. Sleep Med Rev. 2015;29:41–51. [PubMed]

15 

Flygare J, Parthasarathy S, authors. Narcolepsy: let the patient's voice awaken us! Am J Med. 2015;128:10–3. [PubMed Central][PubMed]

16 

Mold JW, Quattlebaum C, Schinnerer E, Boeckman L, Orr W, Hollabaugh K, authors. Identification by primary care clinicians of patients with obstructive sleep apnea: a practice-based research network (PBRN) study. J Am Board Fam Med. 2011;24:138–45. [PubMed]

17 

Netzer NC, Hoegel JJ, Loube D, et al., authors. Prevalence of symptoms and risk of sleep apnea in primary care. Chest. 2003;124:1406–14. [PubMed]

18 

Mold JW, Woolley JH, Nagykaldi Z, authors. Associations between night sweats and other sleep disturbances: an OKPRN study. Ann Fam Med. 2006;4:423–6. [PubMed Central][PubMed]

19 

Haponik EF, Frye AW, Richards B, et al., authors. Sleep history is neglected diagnostic information. Challenges for primary care physicians. J Gen Intern Med. 1996;11:759–61. [PubMed]

20 

Namen AM, Wymer A, Case D, Haponik EF, authors. Performance of sleep histories in an ambulatory medicine clinic: impact of simple chart reminders. Chest. 1999;116:1558–63. [PubMed]

21 

Parthasarathy S, Subramanian S, Quan SF, authors. A multicenter prospective comparative effectiveness study of the effect of physician certification and center accreditation on patient-centered outcomes in obstructive sleep apnea. J Clin Sleep Med. 2014;10:243–9. [PubMed Central][PubMed]

22 

Parthasarathy S, Haynes PL, Budhiraja R, Habib MP, Quan SF, authors. A national survey of the effect of sleep medicine specialists and American Academy of Sleep Medicine accreditation on management of obstructive sleep apnea. J Clin Sleep Med. 2006;2:133–42. [PubMed]

23 

Pamidi S, Knutson KL, Ghods F, Mokhlesi B, authors. The impact of sleep consultation prior to a diagnostic polysomnogram on continuous positive airway pressure adherence. Chest. 2012;141:51–7. [PubMed]