Machine Learning-Based Anomaly Detection Algorithms to Alert Patients Using Sensor Augmented Pump of Infusion Site Failures

Lorenzo Meneghetti, Eyal Dassau, Francis J Doyle 3rd, Simone Del Favero, Lorenzo Meneghetti, Eyal Dassau, Francis J Doyle 3rd, Simone Del Favero

Abstract

Background: Personal insulin pumps have shown to be effective in improving the quality of therapy for people with type 1 diabetes (T1D). However, the safety of this technology is limited by the possible infusion site failures, which are linked with hyperglycemia and ketoacidosis. Thanks to the large availability of collected data provided by modern therapeutic technologies, machine learning algorithms have the potential to provide new way to identify failures early and avert adverse events.

Methods: A clinical dataset (N = 20) is used to evaluate a novel method for detecting real-time infusion site failures using unsupervised anomaly detection algorithms, previously proposed and developed on in-silico data. An adapted feature engineering procedure is introduced to make the method able to operate in the absence of a closed-loop (CL) system and meal announcements.

Results: In the optimal configuration, we obtained a performance of 0.75 Sensitivity (15 out of 20 total failures detected) and 0.08 FP/day, outperforming previously proposed literature algorithms. The algorithm was able to anticipate the replacement of the malfunctioning infusion sets by ~2 h on average.

Conclusions: On the considered dataset, the proposed algorithm showed the potential to improve the safety of patients treated with sensor-augmented pump systems.

Keywords: SAP; anomaly detection; fault detection; infusion site failures; machine learning.

Conflict of interest statement

Declaration of Conflicting Interests: The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: L.M and S.D.F. hold patent applications related to the proposed method. ED is currently an employee and shareholder of Eli Lilly and Company. The work presented in this paper was performed as part of Dr. Dassau’s academic appointment and is independent of his employment with Eli Lilly and Company.

Figures

Figure 1.
Figure 1.
Distribution of failures per patient.
Figure 2.
Figure 2.
Distribution of failures labels.
Figure 3.
Figure 3.
Scheme of the proposed method.
Figure 4.
Figure 4.
Example of the anomaly score (AS) assigned by IF on subject CO-01JG, represented using CGM data, insulin data and infusion set life. The color is given according to the score, normalized between minimum and maximum values observed.
Figure 5.
Figure 5.
Performance obtained by the three anomaly detection algorithms considered and comparison with the performance obtained using the algorithm proposed by Howsmon et al.

References

    1. McAdams BH, Rizvi AA. An overview of insulin pumps and glucose sensors for the generalist. J Clin Med. 2016;5(1):5.
    1. Pickup JC. Management of diabetes mellitus: is the pump mightier than the pen? Nat Rev Endocrinol. 2012;8(7):425-433.
    1. Faradji RN, De la Maza MES. Insulin pump therapy. In: Rodriguez-Saldana Joel. (ed.) The Diabetes Textbook. Springer; 2019:555-570.
    1. Nimri R, Nir J, Phillip M. Insulin pump therapy. Am J Ther. 2020;27(1):e30-e41.
    1. Bergenstal RM, Tamborlane WV, Ahmann A, et al.. Effectiveness of sensor-augmented insulin-pump therapy in type 1 diabetes. N Engl J Med. 2010;363(4):311-320.
    1. Slover RH, Welsh JB, Criego A, et al.. Effectiveness of sensor-augmented pump therapy in children and adolescents with type 1 diabetes in the STAR 3 study. Pediatr Diabetes. 2012;13(1):6-11.
    1. Karlin AW, Ly TT, Pyle L, et al.. Duration of infusion set survival in lipohypertrophy versus nonlipohypertrophied tissue in patients with type 1 diabetes. Diabetes Technol Ther. 2016;18(7):429-435.
    1. Deiss D, Adolfsson P, Alkemade-van Zomeren M, et al.. Insulin infusion set use: European perspectives and recommendations. Diabetes Technol Ther. 2016;18(9):517-524.
    1. Hanas R, Lindgren F, Lindblad B. A 2-yr national population study of pediatric ketoacidosis in Sweden: predisposing conditions and insulin pump use. Pediatr Diabetes. 2009;10(1):33-37.
    1. Dogan ADA, Jørgensen UL, Gjessing HJ. Diabetic ketoacidosis among patients treated with continuous subcutaneous insulin infusion. J Diabetes Sci Technol. 2017;11(3):631-632.
    1. Realsen J, Goettle H, Chase HP. Morbidity and mortality of diabetic ketoacidosis with and without insulin pump care. Diabetes Technol Ther. 2012;14(12):1149-1154.
    1. Maahs DM, Hermann JM, Holman N, et al.. Rates of diabetic ketoacidosis: international comparison with 49,859 pediatric patients with type 1 diabetes from England, Wales, the US, Austria, and Germany. Diabetes Care. 2015;38(10):1876-1882.
    1. Facchinetti A, Favero S, Sparacino G, Cobelli C. An online failure detection method of the glucose sensor-insulin pump system: improved overnight safety of type-1 diabetic subjects. IEEE Trans Biomed Eng. 2013;60(2):406-416.
    1. Del Favero S, Monaro M, Facchinetti A, Tagliavini A, Sparacino G, Cobelli C. Real-time detection of glucose sensor and insulin pump faults in an artificial pancreas. IFAC Proc Vol. 2014;47(3):1941-1946.
    1. Kovacs L, Kulcsar B, Bokor J, Benyo Z. LPV fault detection of glucose-insulin system. In: 14th Mediterranean Conference on Control and Automation, Ancona, Italy, 28-30 June, 2006:1-5.
    1. Vega-Hernandez O, Campos-Cornejo F, Campos-Delgado DU, Espinoza-Trejo DR. Increasing security in an artificial pancreas: diagnosis of actuator faults. 2009. Pan American Health Care Exchanges; 2009:137-142. 10.1109/PAHCE.2009.5158384
    1. Herrero P, Calm R, Vehí J, et al.. Robust fault detection system for insulin pump therapy using continuous glucose monitoring. J Diabetes Sci Technol. 2012;6(5):1131-1141.
    1. Oviedo S, Vehí J, Calm R, Armengol J. A review of personalized blood glucose prediction strategies for T1DM patients. Int J Numer Method Biomed Eng. 2017;33(6):e2833.
    1. Cescon M, DeSalvo DJ, Ly TT, et al.. Early detection of infusion set failure during insulin pump therapy in type 1 diabetes. J Diabetes Sci Technol. 2016;10(6):1268-1276.
    1. Howsmon DP, Cameron F, Baysal N, et al.. Continuous glucose monitoring enables the detection of losses in infusion set actuation (LISAs). Sensors (Basel). 2017;17(1):161.
    1. Howsmon DP, Baysal N, Buckingham BA, et al.. Real-time detection of infusion site failures in a closed-loop artificial pancreas. J Diabetes Sci Technol. 2018;12(3):599-607.
    1. Meneghetti L, Terzi M, Favero SD, Susto GA, Cobelli C. Data-driven anomaly recognition for unsupervised model-free fault detection in artificial pancreas. IEEE Trans Control Syst Technol. 2018;28(1):33-47.
    1. Meneghetti L, Susto GA, Del Favero S. Detection of insulin pump malfunctioning to improve safety in artificial pancreas using unsupervised algorithms. J Diabetes Sci Technol. 2019;13(6):1065-1076.
    1. Goldstein M, Uchida S. A comparative evaluation of unsupervised anomaly detection algorithms for multivariate data. PLoS One. 2016;11(4):e0152173.
    1. Renard E, Farret A, Kropff J, et al.. Day-and-night closed-loop glucose control in patients with type 1 diabetes under free-living conditions: results of a single-arm 1-month experience compared with a previously reported feasibility study of evening and night at home. Diabetes Care. 2016;39(7):1151-1160.
    1. Shalitin S, Ben-Ari T, Yackobovitch-Gavan M, et al.. Using the Internet-based upload blood glucose monitoring and therapy management system in patients with type 1 diabetes. Acta Diabetol. 2014;51(2):247-256.
    1. Saito T, Rehmsmeier M. The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets. PLoS One. 2015;10(3):e0118432.
    1. Liu FT, Ting KM, Zhou Z-H. Isolation forest. In: 2008 Eighth IEEE International Conference on Data Mining; Pisa, Italy, 15-19 December, 2008:413-422.
    1. Goldstein M, Dengel A. Histogram-based outlier score (hbos): A fast unsupervised anomaly detection algorithm. KI 2012 German Conference on Artificial Intelligence. Saarbrücken, Germany, 24-27 September, 2012:59-63.
    1. Ramaswamy S, Rastogi R, Shim K. Efficient algorithms for mining outliers from large data sets. In: Proceedings of the 2000 ACM SIGMOD International Conference on Management of Data Held in Dallas, TX, USA. Association for Computing Machinery; 2000:427-438.
    1. Zhao Y, Nasrullah Z, Li Z. PyOD: a python toolbox for scalable outlier detection. Preprint. Posted online January 6, 2019. arXiv 190101588 [Cs, Stat].
    1. Contreras I, Vehi J. Artificial intelligence for diabetes management and decision support: literature review. J Med Internet Res. 2018;20(5):e10775.
    1. Zimek A, Schubert E, Kriegel H-P. A survey on unsupervised outlier detection in high-dimensional numerical data. Stat Anal Data Min. 2012;5(5):363-387.
    1. Friedman J, Hastie T, Tibshirani R. The Elements of Statistical Learning. vol. 1. Springer Series in Statistics; 2001.

Source: PubMed

Sponsorit ja yhteistyökumppanit

Sairaudet

Huumeiden interventiot

3
Tilaa