Dynamic uncertain causality graph for computer-aided general clinical diagnoses with nasal obstruction as an illustration

Abstract

Many AI systems have been developed for clinical diagnoses, in which most of them lack interpretability in both knowledge representation and inference results. The newly developed Dynamic Uncertain Causality Graph (DUCG) is a probabilistic graphical model with strong interpretability. However, existing DUCG is mainly for fault diagnoses of large, complex industrial systems. In this paper, we extend DUCG for better application in general clinical diagnoses. Four extensions are introduced: (1) special logic gate and zoom function event variables to represent and quantify the influences of various risk factors on the morbidities of diseases. (2) Reversal logic gate to model the case that some diseases/causes may result in at least two simultaneous symptoms/consequences. (3) Disease-specific manifestation variable for special inference and easy understanding to diagnose a specific disease. (4) Event attention importance to count contributions of isolated state-abnormal variables in inference. To illustrate and verify the extended DUCG methodology, we performed a case study for diagnosing 25 diseases causing nasal obstruction. We tested 171 cases randomly selected from total 471 cases of discharged patients in the hospital information system of Xuanwu Hospital. The diagnosis precision of the extended DUCG was 100%. The diagnosis precision of the third-party verification performed by Suining Central Hospital was 98.86%, which exhibited the strong generalization ability of the extended DUCG.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12
Fig. 13
Fig. 14
Fig. 15
Fig. 16
Fig. 17
Fig. 18
Fig. 19
Fig. 20
Fig. 21
Fig. 22

Notes

  1. 1.

    The rules are indexed in series in the DUCG papers.

  2. 2.

    The rules are indexed in series in the DUCG papers.

  3. 3.

    Corollary 15: \(A_{{nk_{n} ;i}} V_{i} A_{{mk_{m} ;i}} V_{i} = \left( {A_{{nk_{n} ;i}} *A_{{mk_{m} ;i}} } \right)V_{i}\), in which

    \(\left( {A_{{nk_{n} ;i}} *A_{{mk_{m} ;i}} } \right) \equiv \left( {\begin{array}{*{20}c} {A_{{nk_{n} ;i1}} A_{{mk_{m} ;i1}} } & {A_{{nk_{n} ;i2}} A_{{mk_{m} ;i2}} } & { \ldots } & {A_{{nk_{n} ;ij}} A_{{mk_{m} ;ij}} } & { \ldots } & {A_{{nk_{n} ;iJ}} A_{{mk_{m} ;iJ}} } \\ \end{array} } \right)\)

    Correspondingly, \(a_{{nk_{n} ;i}} *a_{{mk_{m} ;i}} \equiv \left( {\begin{array}{*{20}c} {a_{{nk_{n} ;i1}} a_{{mk_{m} ;i1}} } & {a_{{nk_{n} ;i2}} a_{{mk_{m} ;i2}} } & { \ldots } & {a_{{nk_{n} ;ij}} a_{{mk_{m} ;ij}} } & { \ldots } & {a_{{nk_{n} ;iJ}} a_{{mk_{m} ;iJ}} } \\ \end{array} } \right)\)

    where, “*” is an AND/multiplication matrix operator specially defined in DUCG. In format, the “*” operator is similar to Hadamard product.

References

  1. Avci E (2011) A new expert system for diagnosis of lung cancer: GDA-LSSVM. J Med Syst 36(3):2005–2009

    Google Scholar 

  2. Bernard O, Lalande A (2018) Deep learning techniques for automatic mri cardiac multi-structures segmentation and diagnosis: is the problem solved? IEEE Trans Med Imaging 37(11):2514–2525

    Google Scholar 

  3. Bhatele KR, Bhadauria SS (2019) Brain structural disorders detection and classification approaches: a review. Artif Intell Rev. https://doi.org/10.1007/s10462-019-09766-9

    Article  Google Scholar 

  4. Brooks R, Heiser J (1980) Some experience with transferring the mycin system to a new domain. IEEE Trans Pattern Anal Mach Intell 2(5):477–478

    Google Scholar 

  5. Chaovalitwongse WA, Pottenger RS et al (2011) Pattern- and network-based classification techniques for multichannel medical data signals to improve brain diagnosis. IEEE Trans Syst Man Cybern A Syst Hum 41(5):977–988

    Google Scholar 

  6. Domingues I, Pereira G, Martins P et al (2019) Using deep learning techniques in medical imaging: a systematic review of applications on CT and PET. Artif Intell Rev. https://doi.org/10.1007/s10462-019-09788-3

    Article  Google Scholar 

  7. Dong C, Zhang Q (2014) Research on weighted logical inference for uncertain fault diagnosis. Chin ACTA Autom Sin 40(12):2766–2781

    MATH  Google Scholar 

  8. Dong C, Wang Y et al (2014) The methodology of dynamic uncertain causality graph for intelligent diagnosis of vertigo. Comput Methods Programs Biomed 133:162–174

    Google Scholar 

  9. Dong C et al (2019) The cubic dynamic uncertain causality graph: a methodology for temporal process modeling and diagnostic logic inference. IEEE Trans Neural Netw Learn Syst. https://doi.org/10.1109/TNNLS.2019.2953177

    Article  Google Scholar 

  10. Dou Q, Chen H et al (2017) Multi-level contextual 3d cnns for false positive reduction in pulmonary nodule detection. IEEE Trans Biomed Eng 63(3):1558–1567

    Google Scholar 

  11. Erickson BJ, Korfiatis P et al (2017) Machine learning for medical imaging. Radiographics 37:505–515

    Google Scholar 

  12. Erickson BJ et al (2018) Deep learning in radiology: does one size fit all? J Am Coll Radiol 15:521–526

    Google Scholar 

  13. Esteva A, Kuprel B et al (2017) Dermatologist-level classification of skin cancer with deep neural networks. Nature 542(7639):115–118

    Google Scholar 

  14. Garg AX, Adhikari NKJ et al (2005) Effects of computerized clinical decision support systems on practitioner performance and patient outcomes: a systematic review. JAMA J Am Med Assoc 280(15):1339–1346

    Google Scholar 

  15. Geng S, Zhang Q (2014) Calculation method to diagnose integrated causes of faults in process system by means of dynamic uncertain causality graph. In: Proceedings of the 2014 Asia-Pacific conference on computer science and applications, Shanghai, China, pp 306–311

  16. Gu Y, Zhang M et al (2019) Fault diagnosis of gearbox based on improved DUCG with combination weighting method. IEEE Access 7:92955–92967

    Google Scholar 

  17. Hao S et al (2017) Intelligent diagnosis of jaundice with dynamic uncertain causality graph model. J Zhejiang Univ Sci B (Biomed Biotechnol) 18(5):393–401

    Google Scholar 

  18. Holt A, Bichindaritz I, Schmidt R, Perner P (2005) Medical applications in case-based reasoning. Knowl Eng Rev 20(3):289–292

    Google Scholar 

  19. Huang CR, Chen YT et al (2016) Gastroesophageal reflux disease diagnosis using hierarchical heterogeneous descriptor fusion support vector machine. IEEE Trans Biomed Eng 63(3):588–599

    Google Scholar 

  20. Huang Q, Chen Y et al (2019) On combining biclustering mining and adaboost for breast tumor classification. IEEE Trans Knowl Data Eng. https://doi.org/10.1109/TKDE.2019.2891622

    Article  Google Scholar 

  21. Iakovidis DK, Georgakopoulos SV et al (2018) Detecting and locating gastrointestinal anomalies using deep learning and iterative cluster unification. IEEE Trans Med Imaging. https://doi.org/10.1109/TMI.2018.2837002

    Article  Google Scholar 

  22. Itani S, Lecron F et al (2018) Specifics of medical data mining for diagnosis aid: a survey. Expert Syst Appl 118:300–314

    Google Scholar 

  23. Jia L, Fang C, Changcun P et al (2018) A cascaded deep convolutional neural network for joint segmentation and genotype prediction of brainstem gliomas. IEEE Trans Biomed Eng 65:1943–1952

    Google Scholar 

  24. Judea P (1988) Probabilistic reasoning in intelligent systems. Morgan Kaufmann, San Mateo CA

    MATH  Google Scholar 

  25. Judea P (2009) Causality: models, reasoning and inference, 2nd edn. Cambridge University Press, New York

    MATH  Google Scholar 

  26. Judea P et al (2018) The book of why—the new science of cause and effect. Hachette, New York

    MATH  Google Scholar 

  27. Keith RDF et al (1996) A muticenter comparative study of 17 experts and an intelligent computer system for managing labor using the cardiotocogram. Int J Gynecol Obstet 53(1):98

    Google Scholar 

  28. Lian C, Liu M, Zhang J et al (2018) Hierarchical fully convolutional network for joint atrophy localization and alzheimer’s disease diagnosis using structural MRI. IEEE Trans Pattern Anal Mach Intell. https://doi.org/10.1109/TPAMI.2018.2889096

    Article  Google Scholar 

  29. Liang H et al (2019) Evaluation and accurate diagnoses of pediatric diseases using artificial intelligence. Nat Madison. https://doi.org/10.1038/s41591-018-0335-9

    Article  Google Scholar 

  30. Lin RH, Chuang CL (2010) A hybrid diagnosis model for determining the types of the liver disease. Comput Biol Med 40(7):665–670

    Google Scholar 

  31. Liu Q (2019) http://nb.ifeng.com/a/20190925/7545950_0.shtml (in Chinese)

  32. Liu X, Chen K et al (2018) Use of multimodality imaging and artificial intelligence for diagnosis and prognosis of early stages of alzheimer’s disease. Transl Res. https://doi.org/10.1016/j.trsl.2018.01.001

    Article  Google Scholar 

  33. Mahfouf M, Abbod MF et al (2001) A survey of fuzzy logic monitoring and control utilisation in medicine. Artif Intell Med 21(1–3):27–42

    Google Scholar 

  34. Marcus G (2018) Deep learning: a critical appraisal. https://arxiv.org/ftp/arxiv/papers/1801/1801.00631.pdf

  35. Markey MK, Lo JY et al (2003) Self-organizing map for cluster analysis of a breast cancer database. Artif Intell Med 27(2):113–127

    Google Scholar 

  36. Meyer AND, Thompson PJ, Khanna A et al (2018) Evaluating a mobile application for improving clinical laboratory test ordering and diagnosis. J Am Med Inf Assoc. https://doi.org/10.1093/jamia/ocy026

    Article  Google Scholar 

  37. Miller RA, Pople HE, Myers JD (1982) Internist-i, an experimental computer-based diagnostic consultant for general internal medicine. N Engl J Med 307(8):468–476

    Google Scholar 

  38. Moghbel M, Ooi CY et al (2019) A review of breast boundary and pectoral muscle segmentation methods in computer-aided detection/diagnosis of breast mammography. Artif Intell Rev. https://doi.org/10.1007/s10462-019-09721-8

    Article  Google Scholar 

  39. Murtaza G, Shuib L et al (2019) Deep learning-based breast cancer classification through medical imaging modalities: state of the art and research challenges. Artif Intell Rev. https://doi.org/10.1007/s10462-019-09716-5

    Article  Google Scholar 

  40. Pal D, Mandana KM et al (2012) Fuzzy expert system approach for coronary artery disease screening using clinical parameters. Knowl Based Syst 36:162–174

    Google Scholar 

  41. Pandey B, Mishra RB (2009) Knowledge and intelligent computing system in medicine. Comput Biol Med 39(3):215–230

    Google Scholar 

  42. Qu Y, Zhang Q et al (2015) Application of dynamic uncertain causality graph to dynamic fault diagnosis in chemical processes. Chin CAAI Trans Intell Syst 10(3):354–361

    Google Scholar 

  43. Rowe SP, Chu LC et al (2019) Computed tomography cinematic rendering in the evaluation of colonic pathology: technique and clinical applications. J Comput Assist Tomogr 43(3):475–484

    Google Scholar 

  44. Ruffle JK, Farmer AD et al (2018) Artificial intelligence-assisted gastroenterology—promises and pitfalls. Am J Gastroenterol. https://doi.org/10.1038/s41395-018-0268-4

    Article  Google Scholar 

  45. Samant P, Agarwal R (2018) Machine learning techniques for medical diagnosis of diabetes using iris images. Comput Methods Progr Biomed. https://doi.org/10.1016/j.cmpb.2018.01.004

    Article  Google Scholar 

  46. Shaban-Nejad A, Michalowski M et al (2018) Health intelligence: how artificial intelligence transforms population and personalized health. npj Digit Med. https://doi.org/10.1038/s41746-018-0058-9

    Article  Google Scholar 

  47. Shortliffe EH, Axline SG et al (1973) An artificial intelligence program to advise physicians regarding antimicrobial therapy. Comput Biomed Res 6(6):544–560

    Google Scholar 

  48. Son LH, Thong NT (2015) Intuitionistic fuzzy recommender systems: an effective tool for medical diagnosis. Knowl Based Syst 74(1):133–150

    Google Scholar 

  49. Vila-Francés Joan et al (2013) Expert system for predicting unstable angina based on bayesian networks. Expert Syst Appl 40(12):5004–5010

    Google Scholar 

  50. Walsh JA, Rozycki M et al (2019) Application of machine learning in the diagnosis of axial spondyloarthritis. Curr Opin Rheumatol. https://doi.org/10.1097/BOR.0000000000000612

    Article  Google Scholar 

  51. Wang F, Zhang P et al (2014) Clinical risk prediction with multilinear sparse logistic regression. In: ACM SIGKDD international conference on knowledge discovery and data mining. ACM

  52. Wu J et al (2018) Master clinical medical knowledge at certificated-doctor-level with deep learning model. Nat Commun. https://doi.org/10.1038/s41467-018-06799-6

    Article  Google Scholar 

  53. Yang Z, Huang Y, Jiang Y et al (2018) Clinical assistant diagnosis for electronic medical record based on convolutional neural network. Sci Rep. https://doi.org/10.1038/s41598-018-24389-w

    Article  Google Scholar 

  54. Yuille AI, Liu C (2019) Deep nets: what have they ever done for vision? IEEE conference on computer vision and pattern recognition, arXiv:1805.04025

  55. Zhang Q (2012) Dynamic uncertain causality graph for knowledge representation and reasoning: discrete DAG cases. J Comput Sci Technol 27:1–23

    MathSciNet  MATH  Google Scholar 

  56. Zhang Q (2015a) Dynamic uncertain causality graph for knowledge representation and probabilistic reasoning: directed cyclic graph and joint probability distribution IEEE Trans. Neural Netw Learn Syst 26:1503–1517

    MathSciNet  Google Scholar 

  57. Zhang Q (2015b) Dynamic uncertain causality graph for knowledge representation and probabilistic reasoning: continuous variable, uncertain evidence and failure forecast. IEEE Trans Syst Man Cybern 45:990–1003

    Google Scholar 

  58. Zhang Q, Geng S (2015) Dynamic uncertain causality graph applied to dynamic fault diagnoses of large and complex systems. IEEE Trans Reliab 64(3):910–927

    Google Scholar 

  59. Zhang Q, Yao Q (2018) Dynamic uncertain causality graph for knowledge representation and reasoning: utilization of statistical data and domain knowledge in complex cases. IEEE Trans Neural Netw Learn Syst 29(5):1637–1651

    MathSciNet  Google Scholar 

  60. Zhang Q, Zhang Z (2015) Dynamic uncertain causality graph applied to dynamic fault diagnoses and predictions with negative feedbacks. IEEE Trans Reliab 65(2):1030–1044

    Google Scholar 

  61. Zhang Q, Dong C et al (2014) Dynamic uncertain causality graph for knowledge representation and probabilistic reasoning: statistics base, matrix, and application. IEEE Trans Neural Netw Learn Syst 25(4):645–663

    Google Scholar 

  62. Zhang Y, Chen M et al (2016) Idoctor: personalized and professionalized medical recommendations based on hybrid matrix factorization. Future Gener Comput Syst 66:30–35

    Google Scholar 

  63. Zhang Q, Qiu K et al (2018) Calculate joint probability distribution of steady directed cyclic graph with local data and domain casual knowledge. China Commun 15:146–155

    Google Scholar 

  64. Zhou Z, Jiang Y (2003) Medical diagnosis with c45 rule preceded by artificial neural network ensemble. IEEE Trans Inf Technol Biomed 7(1):37–42

    Google Scholar 

Download references

Acknowledgements

This research was fully supported by Beijing Tsingrui Intelligence Technology Co., Ltd.

Author information

Affiliations

Authors

Corresponding authors

Correspondence to Qin Zhang or Zhan Zhang.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix

Appendix

Rules 1-10 presented in Zhang (2012) and Rule 16 presented in Zhang (2015a, b), Zhang and Geng (2015), Zhang and Zhang (2015):

  • Rule 1: “If E shows that Zn;i is not met, Fn;i or Pn;i is eliminated from the DUCG. If E shows that Zn;i is met, the conditional Fn;i or Pn;i becomes the ordinary Fn;i or Pn;i.”

  • Rule 2: “If E shows that Vij,V∈{B, X}, is true while Vij is not a parent event of Xn, Fn;i or Pn;i is eliminated from the DUCG.”

  • Rule 3: “If E shows that Xnk is true while Xnk cannot be caused by any states of Vij, V∈{B, X, G}, Fn;i or Pn;i is eliminated from the DUCG, except that Vi is included in a hypothesis, or is a descendant of an event included in a hypothesis, and the causality chain between them is not blocked by an known event.”

  • Rule 4: “If Ε shows that Xnk and Vij, V∈{B, X}, are true while Xnk cannot be caused by Vij, Fn;i or Pn;i is eliminated from the DUCG.”

  • Rule 5: “If the state unknown Xn without input variable or Gn without input variable is encountered, Xn or Gn and its output directed arcs are eliminated from the DUCG.”

  • Rule 6: “If Gi without any output is encountered for any reason, Gi is eliminated from the DUCG.”

  • Rule 7: “If 1) the state of Xn is unknown, 2) Xn does not have any output, and 3) Xn is not predetermined in concern, Xn and all its input directed arcs are eliminated from the DUCG.”

  • Rule 8: If E shows that Xnk and Vij, V∈{B, X}, are true and Xnk appears earlier than Vij, which means that Vij cannot be the cause of Xnk, the F or P type variables (they are the members of the causality chain from Vij to Xnk and are not related to any other upstream causality chain of Xnk) are eliminated from the DUCG.

  • Rule 9: If there is such a group of variables (named as the independent group) that have no causal connection with those variables related to E, and no variable in this group is predetermined in concern, this independent group of variables can be eliminated from the DUCG.

  • Rule 10: If E shows Xnk is true while Xnk does not have any input due to any reason, add a virtual parent event Dn to Xnk with ank;nD = 1 and ank’;nD = 0, k ≠ k’. rn;D can be any value. The added virtual Dn can be drawn as in the simplified graph.”

  • Rule 16: “If (a) E indicates a group of normal state events X,where n∈SI and SI denotes the index set of the variables of this group, (b) X, n∈SI, have no output to other variables, (c) X, n∈SI, are connected with none or a group of state unknown {B-, X-, G-, D-, F-, P-}-type variables given that this group of state unknown variables are isolated by X, n∈SI, and (d) this isolated group of variables are not predetermined in concern, then this isolated group of variables and X, n∈SI, are eliminated from the DUCG, except X that is the descendant of the hypothesis in concern without events in E to block the connection between them.” In this paper, η = 0.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Zhang, Q., Bu, X., Zhang, M. et al. Dynamic uncertain causality graph for computer-aided general clinical diagnoses with nasal obstruction as an illustration. Artif Intell Rev 54, 27–61 (2021). https://doi.org/10.1007/s10462-020-09871-0

Download citation

Keywords

  • Uncertainty
  • Causality
  • Probabilistic reasoning
  • Clinical diagnosis
  • Nasal obstruction