266cb6661fd0f8932201db7c317c5cf6 cancers-17-01101-v4.pdf ba4a9e02d72eff954581e184095ed092303b8ece cancers-17-01101-v4.pdf ebbc12813bf6c7b579a271e7db71c067e449186cdbb30087c3c9ea61c204a243 cancers-17-01101-v4.pdf Title: Developing a Predictive Model for Significant Prostate Cancer Detection in Prostatic Biopsies from Seven Clinical Variables:Is Machine Learning Superior to Logistic Regression? Subject: Objective: This study compares machine learning (ML) and logistic regression (LR) algorithms in developing a predictive model for sPCa using the seven predictive variables from the Barcelona (BCN-MRI) predictive model. Method: A cohort of 5005 men suspected of having PCa who underwent MRI and targeted and/or systematic biopsies was used for training, testing, and validation. A feedforward neural network (FNN)-based SimpleNet model (GMV) and a logistic regression-based model (BCN) were developed. The models were evaluated for discrimination ability, precision–recall, net benefit, and clinical utility. Both models demonstrated strong predictive performance. Results: The GMV model achieved an area under the curve of 0.88 in training and 0.85 in test cohorts (95% CI: 0.83–0.90), while the BCN model reached 0.85 and 0.84 (95% CI: 0.82–0.87), respectively (p > 0.05). The GMV model exhibited higher recall, making it more suitable for clinical scenarios prioritizing sensitivity, whereas the BCN model demonstrated higher precision and specificity, optimizing the reduction of unnecessary biopsies. Both models provided similar clinical benefit over biopsying all men, reducing unnecessary procedures by 27.5–29% and 27–27.5% of prostate biopsies at 95% sensitivity, respectively (p > 0.05). Conclusions: Our findings suggest that both ML and LR models offer high accuracy in sPCa detection, with ML exhibiting superior recall and LR optimizing specificity. These results highlight the need for model selection based on clinical priorities. Keywords: predictive models; prostate cancer detection; machine learning; logistic regression Author: Juan Morote, Berta Miró, Patricia Hernando, Nahuel Paesano, Natàlia Picola, Jesús Muñoz-Rodriguez, Xavier Ruiz-Plazas, Marta V. Muñoz-Rivero, Ana Celma, Gemma García-de Manuel, Pol Servian, José M. Abascal, Enrique Trilla and Olga Méndez Creator: LaTeX with hyperref Producer: pdfTeX-1.40.25 CreationDate: Fri Apr 11 11:34:03 2025 CEST ModDate: Fri Apr 11 11:36:52 2025 CEST Custom Metadata: no Metadata Stream: no Tagged: no UserProperties: no Suspects: no Form: none JavaScript: no Pages: 18 Encrypted: no Page size: 595.276 x 841.89 pts (A4) Page rot: 0 File size: 3417692 bytes Optimized: no PDF version: 1.7 name type encoding emb sub uni object ID ------------------------------------ ----------------- ---------------- --- --- --- --------- WNWTFR+VnURWPalladioL-Bold Type 1 Custom yes yes yes 10 0 SSXPJJ+VnURWPalladioL Type 1 Custom yes yes yes 16 0 AFZWMD+URWPalladioL-Roma Type 1 Custom yes yes yes 21 0 IBROOQ+URWPalladioL-Bold Type 1 Custom yes yes yes 27 0 LHXATD+URWPalladioL-Ital Type 1 Custom yes yes yes 32 0 EILUJF+CMSY10 Type 1 Builtin yes yes yes 61 0 QPPVOU+URWPalladioL-BoldItal Type 1 Custom yes yes yes 72 0 CHHHGA+PalatinoLinotype,Bold TrueType WinAnsi yes yes no 91 0 CHHHGB+PalatinoLinotype CID TrueType Identity-H yes yes yes 94 0 CHHHGC+PalatinoLinotype TrueType WinAnsi yes yes no 100 0 CHHHHD+PalatinoLinotype,Italic TrueType WinAnsi yes yes no 103 0 CHHIFA+PalatinoLinotype TrueType MacRoman yes yes no 106 0 CHHHGA+PalatinoLinotype,Bold TrueType WinAnsi yes yes no 127 0 CHHHGB+PalatinoLinotype CID TrueType Identity-H yes yes yes 130 0 CHHHGC+PalatinoLinotype TrueType WinAnsi yes yes no 136 0 CHHHHD+PalatinoLinotype,Italic TrueType WinAnsi yes yes no 139 0 CHHIFA+PalatinoLinotype TrueType MacRoman yes yes no 142 0 CHHHGA+PalatinoLinotype,Bold TrueType WinAnsi yes yes no 166 0 CHHHGB+PalatinoLinotype CID TrueType Identity-H yes yes yes 169 0 CHHHGC+PalatinoLinotype TrueType WinAnsi yes yes no 175 0 CHHHHD+PalatinoLinotype,Italic TrueType WinAnsi yes yes no 178 0 CHHIFA+PalatinoLinotype TrueType MacRoman yes yes no 181 0 CHHKEJ+PalatinoLinotype,BoldItalic TrueType WinAnsi yes yes no 206 0 CHHHGA+PalatinoLinotype,Bold TrueType WinAnsi yes yes no 209 0 CHHHGB+PalatinoLinotype CID TrueType Identity-H yes yes yes 212 0 CHHHGC+PalatinoLinotype TrueType WinAnsi yes yes no 218 0 CHHHHD+PalatinoLinotype,Italic TrueType WinAnsi yes yes no 221 0 CHHIEP+PalatinoLinotype,Bold TrueType MacRoman yes yes no 224 0 CHHIFA+PalatinoLinotype TrueType MacRoman yes yes no 227 0 CHINFF+Calibri CID TrueType Identity-H yes yes yes 243 0 CHHHGA+PalatinoLinotype,Bold TrueType WinAnsi yes yes no 249 0 CHHHGB+PalatinoLinotype CID TrueType Identity-H yes yes yes 252 0 CHHHGC+PalatinoLinotype TrueType WinAnsi yes yes no 258 0 CHHHHD+PalatinoLinotype,Italic TrueType WinAnsi yes yes no 261 0 CHHIFA+PalatinoLinotype TrueType MacRoman yes yes no 264 0 CHINFF+Calibri CID TrueType Identity-H yes yes yes 278 0 CHHHGA+PalatinoLinotype,Bold TrueType WinAnsi yes yes no 284 0 CHHHGB+PalatinoLinotype CID TrueType Identity-H yes yes yes 287 0 CHHHGC+PalatinoLinotype TrueType WinAnsi yes yes no 293 0 CHHHHD+PalatinoLinotype,Italic TrueType WinAnsi yes yes no 296 0 CHHIFA+PalatinoLinotype TrueType MacRoman yes yes no 299 0 CHJBFN+Arial CID TrueType Identity-H yes yes yes 315 0 CHJGIO+SimSun CID TrueType Identity-H yes yes yes 321 0 CHHHGA+PalatinoLinotype,Bold TrueType WinAnsi yes yes no 327 0 CHHHGB+PalatinoLinotype CID TrueType Identity-H yes yes yes 330 0 CHHHGC+PalatinoLinotype TrueType WinAnsi yes yes no 336 0 CHHHHD+PalatinoLinotype,Italic TrueType WinAnsi yes yes no 339 0 CHHIFA+PalatinoLinotype TrueType MacRoman yes yes no 342 0 Jhove (Rel. 1.28.0, 2023-05-18) Date: 2025-04-23 02:07:53 CEST RepresentationInformation: cancers-17-01101-v4.pdf ReportingModule: PDF-hul, Rel. 1.12.4 (2023-03-16) LastModified: 2025-04-22 09:11:43 CEST Size: 3417692 Format: PDF Version: 1.7 Status: Well-Formed and valid SignatureMatches: PDF-hul MIMEtype: application/pdf PDFMetadata: Objects: 493 FreeObjects: 4 IncrementalUpdates: 0 DocumentCatalog: PageLayout: SinglePage PageMode: UseNone Outlines: Item: Title: Introduction Destination: section.1 Item: Title: Materials and Methods Destination: section.2 Children: Item: Title: Study Design and Participants Destination: subsection.2.1 Item: Title: Diagnostic Approach for Significant Prostate Cancer Destination: subsection.2.2 Item: Title: Predictive Variables Included in the Models and Outcome Variable Destination: subsection.2.3 Item: Title: Algorithms Used for Model Development Destination: subsection.2.4 Item: Title: Statistical Analyses, Algorithm Performance, and Interpretation Destination: subsection.2.5 Item: Title: Results Destination: section.3 Children: Item: Title: Participant Characteristics Destination: subsection.3.1 Item: Title: Model Performance, Calibration, and Validation of the GMV and BCN Predictive Models Destination: subsection.3.2 Item: Title: Variable Importance Interpretation with SHapley Additive exPlanations (SHAP) Destination: subsection.3.3 Item: Title: Clinical Comparison of GMV and BCN Predictive Models for sPCa Detection Destination: subsection.3.4 Item: Title: Discussion Destination: section.4 Item: Title: Conclusions Destination: section.5 Item: Title: Appendix A Destination: appendix.A. Item: Title: References Destination: appendix.B. Info: Title: Developing a Predictive Model for Significant Prostate Cancer Detection in Prostatic Biopsies from Seven Clinical Variables:Is Machine Learning Superior to Logistic Regression? Author: Juan Morote, Berta Miró, Patricia Hernando, Nahuel Paesano, Natàlia Picola, Jesús Muñoz-Rodriguez, Xavier Ruiz-Plazas, Marta V. Muñoz-Rivero, Ana Celma, Gemma García-de Manuel, Pol Servian, José M. Abascal, Enrique Trilla and Olga Méndez Subject: Objective: This study compares machine learning (ML) and logistic regression (LR) algorithms in developing a predictive model for sPCa using the seven predictive variables from the Barcelona (BCN-MRI) predictive model. Method: A cohort of 5005 men suspected of having PCa who underwent MRI and targeted and/or systematic biopsies was used for training, testing, and validation. A feedforward neural network (FNN)-based SimpleNet model (GMV) and a logistic regression-based model (BCN) were developed. The models were evaluated for discrimination ability, precision recall, net benefit, and clinical utility. Both models demonstrated strong predictive performance. Results: The GMV model achieved an area under the curve of 0.88 in training and 0.85 in test cohorts (95% CI: 0.83 0.90), while the BCN model reached 0.85 and 0.84 (95% CI: 0.82 0.87), respectively (p > 0.05). The GMV model exhibited higher recall, making it more suitable for clinical scenarios prioritizing sensitivity, whereas the BCN model demonstrated higher precision and specificity, optimizing the reduction of unnecessary biopsies. Both models provided similar clinical benefit over biopsying all men, reducing unnecessary procedures by 27.5 29% and 27 27.5% of prostate biopsies at 95% sensitivity, respectively (p > 0.05). Conclusions: Our findings suggest that both ML and LR models offer high accuracy in sPCa detection, with ML exhibiting superior recall and LR optimizing specificity. These results highlight the need for model selection based on clinical priorities. Keywords: predictive models; prostate cancer detection; machine learning; logistic regression Creator: LaTeX with hyperref Producer: pdfTeX-1.40.25 CreationDate: Fri Apr 11 11:34:03 CEST 2025 ModDate: Fri Apr 11 11:36:52 CEST 2025 ID: 0xfc5ea7079c60d6301152951e6e28a419, 0xfc5ea7079c60d6301152951e6e28a419 Filters: FilterPipeline: FlateDecode Fonts: Type1: Font: BaseFont: LHXATD+URWPalladioL-Ital FontSubset: true FirstChar: 2 LastChar: 121 FontDescriptor: FontName: LHXATD+URWPalladioL-Ital Flags: Symbolic FontBBox: -170, -305, 1010, 941 FontFile: true EncodingDictionary: Differences: true ToUnicode: true Font: BaseFont: SSXPJJ+VnURWPalladioL FontSubset: true FirstChar: 160 LastChar: 225 FontDescriptor: FontName: SSXPJJ+VnURWPalladioL Flags: Symbolic FontBBox: -166, -283, 1021, 943 FontFile: true EncodingDictionary: Differences: true ToUnicode: true Font: BaseFont: AFZWMD+URWPalladioL-Roma FontSubset: true FirstChar: 2 LastChar: 252 FontDescriptor: FontName: AFZWMD+URWPalladioL-Roma Flags: Symbolic FontBBox: -166, -283, 1021, 943 FontFile: true EncodingDictionary: Differences: true ToUnicode: true Font: BaseFont: QPPVOU+URWPalladioL-BoldItal FontSubset: true FirstChar: 61 LastChar: 112 FontDescriptor: FontName: QPPVOU+URWPalladioL-BoldItal Flags: Symbolic FontBBox: -170, -300, 1073, 935 FontFile: true EncodingDictionary: Differences: true ToUnicode: true Font: BaseFont: WNWTFR+VnURWPalladioL-Bold FontSubset: true FirstChar: 160 LastChar: 160 FontDescriptor: FontName: WNWTFR+VnURWPalladioL-Bold Flags: Symbolic FontBBox: -152, -301, 1000, 935 FontFile: true EncodingDictionary: Differences: true ToUnicode: true Font: BaseFont: IBROOQ+URWPalladioL-Bold FontSubset: true FirstChar: 2 LastChar: 250 FontDescriptor: FontName: IBROOQ+URWPalladioL-Bold Flags: Symbolic FontBBox: -152, -301, 1000, 935 FontFile: true EncodingDictionary: Differences: true ToUnicode: true Font: BaseFont: EILUJF+CMSY10 FontSubset: true FirstChar: 0 LastChar: 21 FontDescriptor: FontName: EILUJF+CMSY10 Flags: Symbolic FontBBox: -29, -960, 1116, 775 FontFile: true ToUnicode: true Pages: Page: Sequence: 1 Annotations: Annotation: Subtype: Link Rect: 34, 177, 152, 186 Annotation: Subtype: Link Rect: 34, 167, 87, 174 Annotation: Subtype: Link Rect: 36, 75, 152, 84 Annotation: Subtype: Link Rect: 34, 63, 89, 72 Annotation: Subtype: Link Rect: 34, 773, 130, 809 Annotation: Subtype: Link Rect: 515, 773, 560, 809 Annotation: Subtype: Link Rect: 192, 643, 203, 656 Annotation: Subtype: Link Rect: 387, 643, 399, 656 Annotation: Subtype: Link Rect: 331, 615, 343, 628 Annotation: Subtype: Link Rect: 409, 27, 560, 37 Annotation: Subtype: Link Rect: 34, 403, 87, 422 Page: Sequence: 2 Annotations: Annotation: Subtype: Link Rect: 416, 510, 423, 519 ActionDest: 347 Annotation: Subtype: Link Rect: 435, 468, 442, 477 ActionDest: 347 Annotation: Subtype: Link Rect: 498, 440, 505, 449 ActionDest: 347 Annotation: Subtype: Link Rect: 360, 329, 367, 338 ActionDest: 347 Annotation: Subtype: Link Rect: 368, 328, 375, 338 ActionDest: 347 Annotation: Subtype: Link Rect: 295, 301, 302, 310 ActionDest: 347 Annotation: Subtype: Link Rect: 302, 301, 309, 310 ActionDest: 347 Annotation: Subtype: Link Rect: 261, 273, 268, 282 ActionDest: 347 Annotation: Subtype: Link Rect: 271, 273, 282, 282 ActionDest: 347 Annotation: Subtype: Link Rect: 407, 189, 419, 198 ActionDest: 347 Annotation: Subtype: Link Rect: 445, 161, 457, 170 ActionDest: 347 Annotation: Subtype: Link Rect: 262, 106, 274, 115 ActionDest: 347 Annotation: Subtype: Link Rect: 275, 106, 287, 115 ActionDest: 347 Page: Sequence: 3 Annotations: Annotation: Subtype: Link Rect: 223, 758, 235, 767 ActionDest: 347 Annotation: Subtype: Link Rect: 323, 731, 335, 740 ActionDest: 347 Annotation: Subtype: Link Rect: 517, 703, 529, 712 ActionDest: 347 Annotation: Subtype: Link Rect: 510, 675, 522, 684 ActionDest: 347 Annotation: Subtype: Link Rect: 525, 675, 537, 684 ActionDest: 350 Annotation: Subtype: Link Rect: 428, 394, 440, 403 ActionDest: 350 Annotation: Subtype: Link Rect: 544, 338, 556, 348 ActionDest: 350 Annotation: Subtype: Link Rect: 234, 311, 246, 320 ActionDest: 350 Annotation: Subtype: Link Rect: 316, 311, 323, 320 ActionDest: 347 Annotation: Subtype: Link Rect: 332, 227, 344, 236 ActionDest: 350 Annotation: Subtype: Link Rect: 393, 199, 405, 208 ActionDest: 350 Annotation: Subtype: Link Rect: 268, 143, 280, 153 ActionDest: 350 Annotation: Subtype: Link Rect: 337, 75, 349, 84 ActionDest: 350 Page: Sequence: 4 Annotations: Annotation: Subtype: Link Rect: 295, 662, 307, 671 ActionDest: 347 Annotation: Subtype: Link Rect: 492, 634, 504, 643 ActionDest: 347 Annotation: Subtype: Link Rect: 346, 353, 361, 365 ActionDest: 196 Annotation: Subtype: Link Rect: 412, 353, 422, 365 ActionDest: 196 Annotation: Subtype: Link Rect: 359, 311, 374, 323 ActionDest: 196 Annotation: Subtype: Link Rect: 544, 79, 556, 88 ActionDest: 350 Page: Sequence: 5 Annotations: Annotation: Subtype: Link Rect: 397, 516, 404, 528 ActionDest: 69 Annotation: Subtype: Link Rect: 465, 376, 480, 388 ActionDest: 196 Page: Sequence: 6 Annotations: Annotation: Subtype: Link Rect: 527, 446, 542, 458 ActionDest: 231 Annotation: Subtype: Link Rect: 267, 405, 273, 417 ActionDest: 77 Annotation: Subtype: Link Rect: 505, 391, 512, 403 ActionDest: 77 Annotation: Subtype: Link Rect: 438, 149, 452, 161 ActionDest: 231 Annotation: Subtype: Link Rect: 477, 94, 484, 106 ActionDest: 111 Page: Sequence: 7 Page: Sequence: 8 Annotations: Annotation: Subtype: Link Rect: 331, 437, 346, 449 ActionDest: 231 Annotation: Subtype: Link Rect: 299, 409, 314, 421 ActionDest: 231 Annotation: Subtype: Link Rect: 472, 381, 487, 393 ActionDest: 231 Annotation: Subtype: Link Rect: 346, 270, 353, 282 ActionDest: 117 Annotation: Subtype: Link Rect: 235, 130, 242, 142 ActionDest: 117 Page: Sequence: 9 Annotations: Annotation: Subtype: Link Rect: 549, 203, 556, 215 ActionDest: 146 Page: Sequence: 10 Annotations: Annotation: Subtype: Link Rect: 310, 509, 317, 521 ActionDest: 146 Annotation: Subtype: Link Rect: 444, 137, 451, 149 ActionDest: 186 Page: Sequence: 11 Annotations: Annotation: Subtype: Link Rect: 399, 282, 411, 291 ActionDest: 350 Annotation: Subtype: Link Rect: 414, 282, 426, 291 ActionDest: 350 Annotation: Subtype: Link Rect: 416, 184, 428, 193 ActionDest: 347 Annotation: Subtype: Link Rect: 429, 184, 441, 193 ActionDest: 347 Annotation: Subtype: Link Rect: 176, 126, 447, 138 Annotation: Subtype: Link Rect: 194, 115, 206, 124 ActionDest: 347 Page: Sequence: 12 Annotations: Annotation: Subtype: Link Rect: 451, 703, 463, 712 ActionDest: 353 Annotation: Subtype: Link Rect: 466, 703, 478, 712 ActionDest: 353 Annotation: Subtype: Link Rect: 544, 508, 556, 517 ActionDest: 353 Annotation: Subtype: Link Rect: 204, 396, 216, 405 ActionDest: 347 Annotation: Subtype: Link Rect: 219, 396, 231, 405 ActionDest: 347 Annotation: Subtype: Link Rect: 318, 368, 330, 377 ActionDest: 347 Annotation: Subtype: Link Rect: 488, 243, 500, 252 ActionDest: 347 Annotation: Subtype: Link Rect: 500, 243, 512, 252 ActionDest: 353 Page: Sequence: 13 Annotations: Annotation: Subtype: Link Rect: 317, 591, 329, 600 ActionDest: 353 Annotation: Subtype: Link Rect: 385, 494, 397, 503 ActionDest: 353 Annotation: Subtype: Link Rect: 382, 466, 394, 475 ActionDest: 347 Annotation: Subtype: Link Rect: 203, 424, 215, 433 ActionDest: 353 Annotation: Subtype: Link Rect: 215, 424, 227, 433 ActionDest: 353 Annotation: Subtype: Link Rect: 526, 326, 560, 337 Annotation: Subtype: Link Rect: 165, 312, 380, 323 Page: Sequence: 14 Page: Sequence: 15 Page: Sequence: 16 Annotations: Annotation: Subtype: Link Rect: 59, 690, 95, 700 Annotation: Subtype: Link Rect: 113, 652, 149, 662 Annotation: Subtype: Link Rect: 122, 613, 158, 624 Annotation: Subtype: Link Rect: 251, 459, 289, 470 Annotation: Subtype: Link Rect: 516, 357, 553, 368 Annotation: Subtype: Link Rect: 256, 306, 292, 317 Annotation: Subtype: Link Rect: 135, 256, 172, 266 Annotation: Subtype: Link Rect: 505, 242, 560, 253 Annotation: Subtype: Link Rect: 56, 230, 145, 241 Annotation: Subtype: Link Rect: 310, 179, 348, 190 Annotation: Subtype: Link Rect: 403, 127, 440, 138 Page: Sequence: 17 Annotations: Annotation: Subtype: Link Rect: 237, 731, 274, 742 Annotation: Subtype: Link Rect: 280, 731, 316, 742 Annotation: Subtype: Link Rect: 519, 705, 557, 716 Annotation: Subtype: Link Rect: 59, 693, 95, 703 Annotation: Subtype: Link Rect: 425, 654, 462, 665 Annotation: Subtype: Link Rect: 261, 628, 298, 639 Annotation: Subtype: Link Rect: 59, 591, 96, 601 Annotation: Subtype: Link Rect: 482, 552, 520, 563 Annotation: Subtype: Link Rect: 211, 514, 249, 524 Annotation: Subtype: Link Rect: 292, 437, 330, 448 Annotation: Subtype: Link Rect: 336, 437, 372, 448 Annotation: Subtype: Link Rect: 443, 412, 481, 422 Annotation: Subtype: Link Rect: 318, 297, 354, 307 Annotation: Subtype: Link Rect: 95, 259, 132, 269 Annotation: Subtype: Link Rect: 162, 220, 200, 231 Annotation: Subtype: Link Rect: 206, 220, 242, 231 Annotation: Subtype: Link Rect: 114, 143, 152, 154 Annotation: Subtype: Link Rect: 307, 105, 345, 116 Page: Sequence: 18 Annotations: Annotation: Subtype: Link Rect: 59, 731, 96, 742 Annotation: Subtype: Link Rect: 350, 705, 388, 716 Annotation: Subtype: Link Rect: 149, 668, 186, 678 Annotation: Subtype: Link Rect: 402, 628, 438, 639 Annotation: Subtype: Link Rect: 521, 603, 557, 614 Annotation: Subtype: Link Rect: 311, 526, 347, 537 Annotation: Subtype: Link Rect: 232, 488, 270, 499 Annotation: Subtype: Link Rect: 152, 450, 190, 461 Checksum: 72fa32f9 Type: CRC32 Checksum: 266cb6661fd0f8932201db7c317c5cf6 Type: MD5 Checksum: ba4a9e02d72eff954581e184095ed092303b8ece Type: SHA-1 Checksum: ebbc12813bf6c7b579a271e7db71c067e449186cdbb30087c3c9ea61c204a243 Type: SHA-256