Skip to main content

Table 2 Decision support systems in urological domain

From: A systematic review of the applications of Expert Systems (ES) and machine learning (ML) in clinical urology

Article

Mdl

Dom

Subdomain

Variables

Output

Knowledge acquisition

Validation method

Target user

[18]

RBR

U Dys

Incontinence in long-term care facilities

Disease related questions

Recommendations

Experts

Comparison to blinded experts and pilot RCT

Non-expert nurses

[15]

RBR

U Dys

U incont treatment

Incontinence symptoms

Behavioural treatment

Agency guidelines

RCT (60) reliability and validity by experts

Patients

[19]

RBR

U Dys

U incont treatment

19 evaluation questionnaires

Individualised health information

An expert and patients’ feedback

No validation

Patients

[20]

RBR

U Dys

U incont

MH, incontinence symptoms, previous incidents and medication history

U incont treatment

Multiple experts, patients record and literature

Evaluation by experts, 95 retrospective data

Non-experts

[16]

RBR

U Dys

Ward management of micturition

LUTS, Urinary tract infection Anatomical obstruction, Multiple causality and sensory impairment

Diagnosis and risk of fall

Multiple experts

Se 0.95, Sp 0.72, Likert scale Cronbach α 0.9

Urology ward nurses

[21]

FRB

U Dys

U dyn interpretation

U dyn variables

Detrusor and sphincter dysfunction

Not mentioned

Improve User Ac by 10%

Experts

[22]

ANN

U Dys

Uroflow interpretation

Value of slopes, frequency and value of maximums, ration of amplitude and total voiding time

Healthy or pathologic Uroflow

Patients data from U dyn

78 test cases ROC 0.7 Ac 79%

Experts

[23]

SVM

U Dys

Diagnosis

Age, examination, Uroflow, U dyn

Healthy or pathologic Uroflow

Patients data

Ac 84%, Se 93%, Sp 33%

Experts

[17]

FNM

U Dys

Diagnosis

46 defining Characteristics from NANDA-I

Diagnosis of U Dys

Multiple experts weighted the variables and literature review

kappa vs experts (0.92–0.42), Se 0.95, Sp 0.92

Experts and non-experts

[14]

FNM

CaP-BPD

Diagnosis of BPE and CaP

Clinical and pathological variables

CaP, BPE medical, BPE surgery

Patients data

10 folds CV AUC 0.86, se 100%, sp 98%

Non-experts

[24]

FRB

CaP-BPD

AP CP CaP BPE

LUTS, quality of life, fever, haematuria, haemospermia, painful ejaculation, fever, perineal pain, bone pain, pyuria, age, DRE

Diagnosis and treatment of prostatic disease

Multiple experts interviews, patients records and literature

Ac 0.76, Se 0.79, Sp 0.75, retrospective data (n = 105)

Residents, patients, medical students

[12]

FRB

CaP-BPD

AP CP CaP BPE

LUTS, quality of life, fever, haematuria, haemospermia, painful ejaculation, fever, perineal pain, bone pain, pyuria, age, DRE

Diagnosis and treatment

WEKA* to extract rules then experts to modify

200 test cases Ac 0.93, Se 0.97, Sp 0.99,

Residents, patients, medical students

[25]

RBR

CaP

Diagnosis before 1st biopsy

Age, race, FH, DRE, PSA, PSAD, PSAV, TRUS findings

Cancer and benign

Not mentioned

25 test cases

Se 100%

Sp 33%

PPV 62%, NPV 100%

Experts

[13]

F-CBR

CaP

Radiotherapy dose for CaP

Gl, PSA, Distribution Volume Histogram

Radiotherapy dose

72 patients’ cases

Comparison to experts, Ac 85%

Experts

[26]

F-ONT

BPD

Diagnosis and treatment of BPE

LUTS, DRE

Watchful waiting, medical, surgery

Multiple experts weighted the variables

44 prospective cases, agreement kappa = 0.89

Experts and non-experts

[27]

RBR

S Dys

Diagnosis and treatment

Set of descriptors

Therapeutic dialogue

Not mentioned

10 Patients' evaluations

Couples

[28]

RBR

S Dys

Male S dys diagnosis

22 parameters from history and examination

ED diagnosis

GA rule extraction from 30 cases

Se (73–94%), Sp (78–96%)

Ac (89%) vs Residents

Un specified

[29]

FRB

S Dys

Male S dys diagnosis and treatment

MH, non-coital erection, diabetes mellitus, coronary artery, neuropathies, sexual history, psychosocial history, depression, smoking, alcohol, examination, hormonal evaluation, cholesterol

Diagnosis and treatment of ED

Multiple experts’ interviews, Pearson analysis on variables from patients' data and literature

70 test cases vs experts and non-experts (Ac79%)

Non-experts

[30]

FNM

UTI

UTI treatment

Clinical data on UTI

Antibiotics course

Patients data and guidelines

Ac 86.8%, 38 random cases

Experts and non-experts

[31]

ANN

VUR

Decision support for intervention

Age, gender, number of UTIs prior to VUR diagnosis, UTI, of complete ureteral duplication noted on Ultrasound, the presence of bowel or bladder dysfunction

UTI or not

255 cases, 96 cases

AUC 0.76

Experts

[32]

ANN

Nlt

ESWL dose calculation

Age, stone size, stone burden, number of sittings

Number and power of shock

196 cases, 80 cases

coefficient of correlation 0.9

Experts

  1. A total of 21 Expert Systems included supporting the decision making in Urological domains. Rule based reasoning was the most common model and urinary dysfunction was the commonest domain