An Intelligent System for Diagnostics of Pancreatic Diseases

Abstract

This paper describes an intelligent system that performs the JSM automated research support method, which is designed to diagnose pancreatic diseases, that is, chronic pancreatitis and pancreatic cancer. A preliminary study is presented; further trends for the development of the system are listed.

Appendices

APPENDIX 1

The list of features used in the fact base and their data types

No.	Sign name	Data type
	1. Clinical data
1	1.1 Gender	Enumeration
2	1.2 Age	Integer
3	1.3 Body mass index	Number with two decimal digits
4	1.4 Duration of the disease	Number with two decimal digits
5	1.5 Presence of alcohol addiction	Binary type
6	1.6 Availability of tobacco addiction	Binary type
7	1.7 Development of diabetes	Binary type
8	1.8 Pancreatic cancer	Binary type
	2. Laboratory data
	2.1 Biochemistry
9	2.1.1 Total bilirubin	Number with two decimal digits
10	2.1.2 Direct bilirubin	Number with two decimal digits
11	2.1.3 Indirect bilirubin	Number with two decimal digits
12	2.1.4 Gamma-glutamyltranspeptidase (GGTP)	Number with two decimal digits
13	2.1.6 Glucose	Number with two decimal digits
14	2.1.5 Total protein	Number with two decimal digits
15	2.1.7 c-peptide	Number with two decimal digits
16	2.1.8 Fecal elastase	Number with two decimal digits
	2.2 General blood test
17	2.2.1 Hemoglobin	Number with two decimal digits
18	2.2.2 White blood cells	Number with two decimal digits
19	2.2.3 ESR	Number with two decimal digits
	2.3 Oncomarkers
20	2.3.1 CA 19-9	Number with two decimal digits
21	2.3.2 CA 242	Number with two decimal digits
22	2.3.3 CEA	Number with two decimal digits
	3. Ultrasound examination
	3.1 Reliable signs of pancreatic cancer (PC)
23	3.1.1 Identification of a volumetric neoplasm (more often solid), hypo and isoechoic identification	Binary type
	3.2 Indirect signs of PC
24	3.2.1 Uniform dilatation of the main pancreatic duct (MPD) without pronounced wall compaction	Binary type
	3.3 Direct signs of chronic pancreatitis
25	3.3.1 Calcifications	Binary type
	3.4 Signs of chronic pancreatitis
26	3.4.1 Hyperechoic structure of the pancreas	Binary type
27	3.4.2 Uneven dilatation of the MPD, compaction of its walls	Binary type
	4. Computed tomography (CT)
28	4.1 Neoplasms in the structure of the pancreas	Binary type
29	4.2 Biliary hypertension	Binary type
	4.3 Dilatation of the MPD
30	4.3.1 No	Binary type
31	4.3.2 Yes, regular	Binary type
32	4.3.3 Yes, irregular	Binary type
	4.4 Densitometric characteristics for pancreatic cancer in phases, HU
	4.4.1 Native
33	4.4.1.1 min	Integer
34	4.4.1.1 max	Integer
	4.4.2 Arterial
35	4.4.2.1 min	Integer
36	4.4.2.1 max	Integer
	4.4.3 Venous
37	4.4.3.1 min	Integer
38	4.4.3.1 max	Integer
	4.4.4 Delayed
39	4.4.4.1 min	Integer
40	4.4.4.1 max	Integer
	4.5 Density gradient between tumor and unchanged tissue
	4.5.1 Native
41	4.5.1.1 min	Integer
42	4.5.1.1 max	Integer
	4.5.2 Arterial
43	4.5.2.1 min	Integer
44	4.5.2.1 max	Integer
	4.5.3 Venous
45	4.5.3.1 min	Integer
46	4.5.3.1 max	Integer
	4.5.4 Delayed
47	4.5.4.1 min	Integer
48	4.5.4.1 max	Integer
	4.6 Gradient average
49	4.6.1 Native	Number with two decimal digits
50	4.6.2 Arterial	Number with two decimal digits
51	4.6.3 Venous	Number with two decimal digits
52	4.6.4 Delayed	Number with two decimal digits

An example of a correct prediction

The source example for prediction:

0—Id	73
1—Gender	M
2—Age	72
3—BMI	26.4
4—Duration of the disease	4
5—Alcohol	Yes
6—Smoking	Yes
7—ID	No
8—Pancreatic cancer
9—Total bilirubin	12.9
10—Direct bilirubin	2.9
11– Indirect bilirubin	10
12– GGTP	202
13—Total protein	70.7
14—Glucose	5.9
15—C-peptide	0.6
16—Fecal elastase	296
17—Hemoglobin	127
18—White blood cells	6.3
19—ESR	33
20—CaA19-9	974
21—CA 242	150
22—CEA	4.5
23—Detection of a voluminous neoplasm (more often solid), hypo and isoechoic detection	Yes
24—(US) Regular dilatation of the MPD without pronounced compaction of its walls	Yes
25—(US) Regular dilatation of the MPD without pronounced compaction of its walls	No
26—Hyperechoic structure of the pancreas	Yes
27—(US) Irregular dilatation of the MPD, compaction of its walls	No
28—Neoplasms in the structure of the pancreas	Yes
29—Biliary hypertension	Yes
30—(CT) There is a dilatation of the MPD	No
31—(CT) Regular dilatation of the MPD	Yes
32—(CT) Irregular dilatation of the MPD	No
33—Densitometry (native, min)	20
34—Densitometry (native, max)	77
35—Densitometry (arterial, min)	16
36—Densitometry (arterial, max)	106
37—Densitometry (venous, min)	24
38—Densitometry (venous, max)	121
39—Densitometry (delayed, min)	37
40—Densitometry (delayed, max)	137
41– Gradient (native, min)	24
42—Gradient (native, max)	8
43—Gradient (arterial, min)	8
44—Gradient (arterial, max)	6
45—Gradient (venous, min)	43
46—Gradient (venous, max)	23
47—Gradient (delayed, min)	22
48—Gradient (delayed, max)	29
49—Gradient average (native)	16
50—Gradient average (arterial)	7
51—Gradient average (venous)	33
52—Gradient average (delayed)	25.5
In this example, the system diagnosed PC using the following hypotheses (the signs that have no values in the hypothesis are omitted):
Hypothesis 1
22—CEA	3.1–10.2
25—(US) Regular dilatation of the MPD without pronounced compaction of walls	No
26—Hyperechoic structure of the gland	Yes
28—Neoplasms in the structure of the pancreas	Yes
49—Gradient average (native)	9–16
Hypothesis 2
4—Duration of the disease	0.45–4
14—Glucose	5.13–6.27
23—Detection of a voluminous neoplasm (more often solid), hypo and isoechoic detection	Yes
25—US) Regular dilatation of the MPD without pronounced compaction of walls	No
26—Hyperechoic structure of the gland	Yes