Abstract
Resistance to adopting machine-learning algorithms in clinical practice may be due to a perception that these are “black-box” techniques and incompatible with decision-making based on evidence and clinical experience. We believe this resistance is unfortunate, given the increasing availability of large databases containing assessment information that could benefit from machine-learning and data-mining techniques, thereby providing a new and important source of evidence upon which to base clinical decisions. We have focused our investigation on the clinical applications of machine-learning algorithms on older persons in a home care rehabilitation setting. Data for this research were obtained from standardized client assessments using the comprehensive RAI-Home Care (RAI-HC) assessment instrument. Our work has shown that machine-learning algorithms can produce better decisions than standard clinical protocols. More importantly, we have shown that machine-learning algorithms can do much more than make “black-box” predictions; they can generate important new clinical and scientific insights. These insights can be used to make better decisions about treatment plans for patients and about resource allocation for healthcare services, resulting in better outcomes for patients, and in a more efficient and effective healthcare system.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
References
Akaike H (1974) A new look at the statistical model identification. IEEE T Automat Contr 19:716–723
Armstrong JJ, Stolee P, Hirdes JP, Poss JW (2010) Examining three frailty conceptualizations in their ability to predict negative outcomes for home care clients. Age Ageing 39:755–758
Bayes T (1763) An essay towards solving a problem in the doctrine of chances. Phil Trans Roy Soc 53:370–418
Begg R, Kamruzzaman J (2005) A machine learning approach for automated recognition of movement patterns using basic, kinetic and kinematic gait data. J Biomech 38:401–408
Bishop CM (2006) Pattern recognition and machine learning. Springer, New York
Boser B, Guyon I, Vapnik VN (1992) A training algorithm for optimal margin classifiers. In: Proceeding 5th annual workshop computer learn theory. ACM Press, New York
Borrie MJ, Stolee P, Knoefel FD, Wells JL, Seabrook JA (2005) Current best practices in geriatric rehabilitation in Canada. Geriatr Today Can J Geriatr Med Geriatr Psych 8:148–153
Breiman L (1996) Bagging predictors. Mach Learn 24:123–140
Breiman L (2001) Random forests. Mach Learn 45:5–32
Breiman L (2001) Statistical modeling: the two cultures. Stat Sci 16:199–231
Colombo M, Guaita A, Cottino M, Previderé G, Ferrari D, Vitali S (2004) The impact of cognitive impairment on the rehabilitation process in geriatrics. Arch Gerontol Geriatr Suppl 9:85–92
Cover TM, Hart PE (1967) Nearest neighbor pattern classification. IEEE T Inform Theory 13:21–27
Cristianini N, Shawe-Taylor J (2002) An introduction to Support Vector Machines and Other Kernel-Based Learning Methods. Cambridge University Press, New York
Diwan S, Shugarman LR, Fries BE (2004) Problem identification and care plan responses in a home and community-based services program. Med Care 23:193–211
Efron B (1979) Bootstrap methods: another look at the jackknife. Ann Stat 7:1–26
Efron B, Hastie TJ, Johnstone IM, Tibshirani RJ (2004) Least angle regression (with discussion). Ann Stat 32:407–499
Fan J, Li R (2001) Variable selection via nonconcave penalized likelihood and its oracle properties. J Amer Statist Assoc 96:1348–1360
Freund Y, Schapire R (1996) Experiments with a new boosting algorithm. Machine Learning Proc 13th Int Conf. Morgan Kauffman, San Francisco, pp 148–156
Friedman JH (2001) Greedy function approximation: a gradient boosting machine. Ann Stat 29:1189–1232
Friedman JH, Hastie TJ, Tibshirani RJ (2000) Additive logistic regression: A statistical view of boosting. Ann Stat 28:337–407
Ghisla MK, Cossi S, Timpini A, Baroni F, Facchi E, Marengoni A (2007) Predictors of successful rehabilitation in geriatric patients: subgroup analysis of patients with cognitive impairment. Aging Clin Exp Res 19:417–423
Gill PE, Murray W, Wright MH (1986) Practical optimization. Academic Press, London
Gilmore S, Hofmann-Wellenhof R, Soyer HP (2010) A support vector machine for decision support in melanoma recognition. Exp Dermatol 19:830–835
Hanks RA, Rapport LJ, Millis SR, Deshpande SA (1999) Measures of executive functioning as predictors of functional ability and social integration in a rehabilitation sample. Arch Phys Med Rehabil 80:1030–1037
Harrison RF, Kennedy RL (2005) Artificial neural network models for prediction of acute coronary syndromes using clinical data from the time of presentation. Ann Emerg Med 46:431–439
Hastie TJ, Tibshirani RJ, Friedman JF (2001) The elements of statistical learning: data mining, Inference and Prediction. Springer, New York
Hepburn B (2010) Healthcare in Ontario is cracking under stress. Toronto Star, August
Hershkovitz A, Kalandariov Z, Hermush V, Weiss R, Brill S (2007) Factors affecting short-term rehabilitation outcomes of disabled elderly patients with proximal hip fracture. Arch Phys Med Rehabil 88:916–921
Hirdes JP, Fries BE, Morris JN, Steel K, Mor V, Frijters DH et al (1999) Integrated health information systems based on the RAI/MDS series of instruments. Health Manage Forum 12:30–40
Hirdes JP, Frijters D, Teare G (2003) The MDS-CHESS scale: a new measure to predict mortality in institutionalized older people. J Am Geriatr Soc 51:96–100
Hirdes JP, Fries BE, Morris JN, Ikegami N, Zimmerman D, Dalby DM et al (2004) Home Care Quality Indicators (HCQIs) based on the MDS-HC. Gerontologist 44:665–679
Hirdes JP, Poss JW, Curtin-Telegdi N (2008) The method for assigning priority levels (MAPLe): a new decision-support system for allocating home care resources. BMC Med 6:9
Ji SY, Smith R, Huynh T, Najarian K (2009) A comparative analysis of multi-level computer-assisted decision-making systems for traumatic injuries. BMC Med Inform Decis Mak 9:2
Kulminski AM, Ukraintseva SV, Kulminskaya IV, Arbeev KG, Land K, Yashin AI (2008) Cumulative deficits better characterize susceptibility to death in elderly people than phenotypic frailty: lessons from the cardiovascular health study. J Am Geriatr Soc 56:898–903
Lau HY, Tong KY, Zhu H (2009) Support vector machine for classification of walking conditions of persons after stroke with dropped foot. Hum Mov Sci 28:504–514
Landi F, Tua E, Onder G, Carrara B, Sgadari A, Rinaldi C et al (2000) Minimum data set for home care: a valid instrument to assess frail older people living in the community. Med Care 38:1184–1190
Lucas P (2004) Bayesian analysis, pattern analysis, and data mining in healthcare. Curr Opin Crit Care 10:399–403
McCullagh P, Nelder JA (1989) Generalized Linear Models. Chapman and Hall, Boca Raton
Meier L, van de Geer S, Bühlmann P (2008) The group Lasso for logistic regression. J Royal Stat Soc B Met 70:53–71
Meinshausen N (2007) Relaxed Lasso. Comput Stat Data An 52:374–393
Meinshausen N, Bühlmann P (2010) Stability selection (with discussion). J Royal Stat Soc B Met 72:417–473
Melin R, Fugl-Meyer AR (2003) On prediction of vocational rehabilitation outcome at a Swedish employability institute. J Rehabil Med 35:284–289
Miller AJ (2002) Subset selection in regression. Chapman and Hall, New York
Mitnitski AB, Graham JE, Mogilner AJ, Rockwood K (2002) Frailty, fitness and late-life mortality in relation to chronological and biological age. BMC Geriatr 2:1
Mitnitski AB, Mogilner AJ, Graham JE, Rockwood K (2003) Techniques for knowledge discovery in existing biomedical databases: Estimation of individual aging effects in cognition in relation to dementia. J Clinical Epidemiol 56:116–123
Morris JN, Fries BE, Mehr DR, Hawes C, Phillips C, Mor V et al (1994) MDS cognitive performance scale. J Gerontol 49:M174–M182
Morris JN, Fries BE, Morris SA (1999) Scaling ADLs within the MDS. J Gerontol Med Sci 54:M546–M553
Morris JN, Fries BE, Steel K, Ikegami N, Bernabei R, Carpenter GI et al (1997) Comprehensive clinical assessment in community setting: applicability of the MDS-HC. J Am Geriatr Soc 45:1017–1024
Morris JN, Fries BE, Steel K, Ikegami N, Bernabei R (1999) Primer on use of the Minimum Data Set-Home Care (MDS-HC) version 2.0© and the Client Assessment Protocols (CAPs). Hebrew Rehabilitation Center for Aged, Boston
Naglie G, Tansey C, Kirkland JL, Ogilvie-Harris DJ, Detsky AS, Etchells E, Tomlinson G, O’Rourke K, Goldlist B (2002) Interdisciplinary inpatient care for elderly people with hip fracture: a randomized controlled trial. Can Med Assoc J 167:25–32
Ottenbacher KJ, Linn RT, Smith PM, Illig SB, Mancuso M, Granger CV (2004) Comparison of logistic regression and neural network analysis applied to predicting living setting after hip fracture. Ann Epidemiol 14:551–559
Pearce CB, Gunn SR, Ahmed A, Johnson CD (2006) Machine learning can improve prediction of severity in acute pancreatitis using admission values of APACHE II score and C-reactive protein. Pancreatology 6:123–131
Pepe MS (2003) The statistical evaluation of medical tests for classification and prediction. Oxford University Press, New York
Price RK, Spitznagel EL, Downey TJ, Meyer DJ, Risk NK, El-Ghassawy OG (2000) Applying artificial neural network models to clinical decision making. Psychol Assess 12:40–51
Radchenko P, James G (2008) Variable inclusion and shrinkage algorithms. J Amer Statist Assoc 103:1304–1315
Ramos-Pollán R, Guevara-López MA, Suárez-Ortega C, Díaz-Herrero G, Franco-Valiente JM, Rubio-Del-Solar M, González-de-Posada N, Vaz MA, Loureiro J, Ramos I (2012) Discovering mammography-based machine learning classifiers for breast cancer diagnosis. J Med Syst 36(4):2259–2269
Rockwood K, Mitnitski AB (2007) Frailty in relation to the accumulation of deficits. J Gerontol A Biol Sci Med Sci 62:722–727
Shmueli G (2010) To explain or to predict? Stat Sci 25:289–310
Searle S, Mitnitski AB, Gahbauer E, Gill T, Rockwood K (2008) A standard procedure for creating a frailty index. BMC Geriatr 8:24
Stolee P, Borrie MJ, Cook S, Hollomby J, the participants of the Canadian Consensus Workshop on Geriatric Rehabilitation (2004) A research agenda for geriatric rehabilitation: the Canadian consensus. Geriatr Today J Can Geriatr Soc 7:38–42
Tam SF, Cheing GLY, Hui-Chan SWY (2004) Predicting osteoarthritic knee rehabilitation outcome by using a prediction model using data mining techniques. Int J Rehabil Res 27:65–69
Thorsen L, Gjerset GM, Loge JH, Kiserud CE, Skovlund E, Fløtten T, Fosså SD (2011) Cancer patients’ needs for rehabilitation services. Acta Oncol 50:212–222
Tibshirani R (1996) Regression shrinkage and selection via the Lasso. J Royal Stat Soc B 58:267–288
Vapnik VN (1995) The nature of statistical learning theory. Springer, New York
Wells JL, Seabrook JA, Stolee P, Borrie MJ, Knoefel F (2003) State of the art in geriatric rehabilitation. Part I: review of frailty and comprehensive geriatric assessment. Arch Phys Med Rehabil 84:890–897
Wells JL, Seabrook JA, Stolee P, Borrie MJ, Knoefel F (2003) State of the art in geriatric rehabilitation. Part II: clinical challenges. Arch Phys Med Rehabil 84:898–903
Williams AP, Lum JM, Deber R, Montgomery R, Kuluski K, Peckham A et al (2009) Aging at home: integrating community-based care for older persons. Healthc Pap 10:8–21
Xin L, Zhu M (2012) Stochastic stepwise ensembles for variable selection. J Comput Graph Stat 21:275–294
Zou H (2006) The adaptive Lasso and its oracle properties. J Am Stat Assoc 101:1418–1429
Zou H, Hastie TJ (2005) Regularization and variable selection via the elastic net. J Royal Stat Soc B 67:301–320
Zhu M (2008) Kernels and ensembles: perspectives on statistical learning. Am Stat 62:97–109
Zhu M, Chipman HA (2006) Darwinian evolution in parallel universes: a parallel genetic algorithm for variable selection. Technometrics 48:491–502
Zhu M, Chen W, Hirdes JP, Stolee P (2007) The K-nearest neighbors algorithm predicted rehabilitation potential better than current clinical assessment protocol. J Clin Epidemiol 60:1015–1021
Zhu M, Zhang Z, Hirdes JP, Stolee P (2007) Using machine learning algorithms to guide rehabilitation planning for home care clients. BMC Med Inform Dec Mak 7:41
Acknowledgments
The InfoRehab project is supported by the Canadian Institutes of Health Research (CIHR). We thank Chloe Wu for her assistance with the management of data.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendix: Evaluation of Binary Predictions
Appendix: Evaluation of Binary Predictions
Suppose we have a certain procedure (whether an algorithm or a protocol) for predicting binary outcomes of either zero or one. The false positive (FP) and false negative (FN) rates are intuitive measures of the prediction performance. They are the probabilities of the two types of errors the procedure can make, namely, calling a true zero a one (FP) and calling a true one a zero (FN).
The positive diagnostic likelihood ratio \( (DLR + ) \), and the negative diagnostic likelihood ratio \( (DLR - ) \) are less intuitive but extremely useful measures; they
“quantify the change in the odds of [the true outcome] obtained by knowledge of [the prediction]” or “the increase in knowledge about [the true outcome] gained through [the prediction]” [53].
Let
By a simple application of Bayes’ theorem [3], it can be shown that
Therefore, \( DLR + \) can be interpreted as the factor by which a prediction of one can increase the prior odds, and \( DLR{-} \) can be interpreted as the factor by which a prediction of zero can decrease the prior odds. Therefore, informative prediction procedures should have \( DLR{+} > 1 \) and \( DLR\,- < 1 \). Given two prediction methods, A and B, A can be said to be more informative than B if \( DLR{+} (A)\, > DLR + (B) \) and if \( DLR{-} (A) < DLR - (B) \). The use of \( DLR + \) and \( DLR - \) to evaluate procedures for making binary predictions has been gaining popularity in the last two decades [53].
Rights and permissions
Copyright information
© 2014 Springer-Verlag Berlin Heidelberg
About this chapter
Cite this chapter
Zhu, M., Cheng, L., Armstrong, J.J., Poss, J.W., Hirdes, J.P., Stolee, P. (2014). Using Machine Learning to Plan Rehabilitation for Home Care Clients: Beyond “Black-Box” Predictions. In: Dua, S., Acharya, U., Dua, P. (eds) Machine Learning in Healthcare Informatics. Intelligent Systems Reference Library, vol 56. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-40017-9_9
Download citation
DOI: https://doi.org/10.1007/978-3-642-40017-9_9
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-40016-2
Online ISBN: 978-3-642-40017-9
eBook Packages: EngineeringEngineering (R0)