Original article
Robotic-assisted vs. open radical prostatectomy: A machine learning framework for intelligent analysis of patient-reported outcomes from online cancer support groups

https://doi.org/10.1016/j.urolonc.2018.08.012Get rights and content

Highlights

  • •

    PRIME-2 framework (Patient Reported Information Multidimensional Exploration version 2) is a machine learning platform which provides a novel method of intelligent analysis of online cancer support groups.

  • •

    This framework was used to identify decision factors, functional and emotional outcomes of patients undergoing robotic (RALP) vs. open radical prostatectomy (ORP).

  • •

    ORP and RARP demonstrated similar side effect profiles at 12 months, but PRIME-2 enables identification of important QOL features and emotions over time.

Abstract

Background

The advantages of Robot-assisted laparoscopic prostatectomy (RARP) over open radical prostatectomy (ORP) in Prostate cancer perioperatively are well-established, but quality of life is more contentious. Increasingly, patients are utilising online cancer support groups (OCSG) to express themselves. Currently there is no method of analysis of these sophisticated data sources. We have used the PRIME-2 (Patient Reported Information Multidimensional Exploration version 2) framework for automated identification and intelligent analysis of decision-making, functional and emotional outcomes in men undergoing ORP vs. RARP from OCSG discussions.

Methods

The PRIME-2 framework was developed to retrospectively analyse individualised patient-reported information from 5,157 patients undergoing RARP and 579 ORP. The decision factors, side effects, and emotions in 2 groups were analysed and compared using Chi-squared, t tests, and Pearson correlation.

Results

There were no differences in Gleason score, Prostate Specific Antigen (PSA), and age between the groups. Surgeon experience and preservation of erectile function (P < 0.01) were important factors in the decision making process.

There were no significant differences in urinary, sexual, or bowel symptoms between ORP and RARP on a monthly basis during the initial 12 months. Emotions expressed by patients undergoing RARP were more consistent and positive while ORP expressed more negative emotions at the time of surgery and 3 months postsurgery (P < 0.05), due to pain and discomfort, and during ninth month due to fear and anxiety of pending PSA tests.

Conclusions

ORP and RARP demonstrated similar side effect profiles for 12 months, but PRIME-2 enables identification of important quality of life features and emotions over time. It is timely for clinicians to accept OCSG as an adjunct to Prostate cancer care.

Introduction

Robot-assisted radical prostatectomy (RARP) is being increasingly utilised in treatment of prostate cancer (CaP). In addition to the minimally invasive nature, some of the reported perioperative advantages of RARP include lower surgical margin rate for intermediate-risk disease and high-risk disease, lower blood loss, lower risk of blood transfusions, and shorter hospital stays [1], [2], [3], [4], [5], [6], [7]. However, quality of life (QOL) in the longer term is more contentious. Clinical trials conducted in high volume centres depend on willing participants filling out questionnaires on a long-term basis in a ‘trial setting’ and thereby likely to be impacted by a variety of follow-up biases and may not accurately capture ‘real-life’ issues experienced by patients. While main strength in the use of validated questionnaires is standardised, accurate, quantitative comparison between groups, patient-level qualitative expressions of symptomology, and anxiety which is important to the individual may not be accurately captured.

Due to increased prevalence of technology, many patients utilise online cancer support groups (OCSG), to search for information, share experiences, and discuss treatment options [8]. OCSG contain an untapped wealth of implicit patient-reported information encapsulated within voluminous bodies of unstructured text data, providing access to large numbers of patients. Currently there is no recognised method of assessing this individual patient-level data on a large scale.

In previous work, we developed the Patient Reported Information Multidimensional Exploration (PRIME) framework [9], which was demonstrated and validated using a cohort of low-intermediate risk CaP patients undergoing treatment [10], [11]. We have extended and advanced the PRIME framework [9], [10], [11], as PRIME-2, with new machine learning and deep learning algorithms, for automatic extraction and time-based analysis of individual patient trajectories, decisions, outcomes and emotions, superseding the need for patients to fill out questionnaires to collect this information. In this study, we have used PRIME-2 for automated intelligent analysis of decision-making, functional and emotional outcomes in men undergoing ORP vs. RARP from OCSG discussions.

Section snippets

Study population and inclusion criteria

We retrospectively analysed 10 active, high-volume international OCSG focused on CaP discussions (Appendix table S1). An active, high-volume OCSG is defined by having at least 100 new conversations per week. From among these active OCSG, conversations were automatically filtered using the specific topic ‘prostate cancer’. Thereby, 10 active CaP-related OCSG are filtered without any manual searching or intervention. Conversations on OCSG are publicly available data, however, we have obtained

Patient characteristics

5,157 men who decided to undergo RARP were compared with 579 ORP over a period of 15 months, from 3 months prior to their decision to 12 months folllowing their treatment.

Table 1 demonstrates patient characteristics RARP vs. ORP. There were no differences in gleason score, PSA, age, urinary, or sexual symptoms between the groups.

Key decision factors for RARP vs. ORP

The main factors influencing the decision making process between RARP vs. ORP were surgeon skill/experience (92% vs. 89%), preservation of erectile function (53% vs.

Discussion

There is an increasing trend towards the utility of RARP over the last decade [12] and has shown to provide optimism for men who wanted surgery for CaP, but feared the morbidity associated with traditional open surgery [13]. However, the decision to choose between ORP and RARP is complex and is influenced by many factors including surgeon's experience, availability of resources, patient's expectations of the procedure and recovery period [14], [15]. In different international healthcare

Conclusion

Based on the previously validated PRIME framework, we developed PRIME-2, to analyse and validate outcomes of men undergoing RARP vs. ORP. There were no differences in urinary, sexual, or bowel function between the 2 groups. However, there were differences in patient emotions. As a first-time study of patient reported emotions over a period of 12 months, it is important to note the fluctuations and intensities at different points in time. Intense expression of negative emotions postsurgery by

Acknowledgement

We would like to acknowledge Dr Luke Wang for his administrative support

References (35)

  • E. Haglind et al.

    Urinary incontinence and erectile dysfunction after robotic versus open radical prostatectomy: a prospective, controlled, nonrandomised trial

    Eur Urol

    (2015)
  • G. Gandaglia et al.

    Comparative effectiveness of robot-assisted and open radical prostatectomy in the postdissemination era

    J Clin Oncol

    (2014)
  • A. Tewari et al.

    A prospective comparison of radical retropubic and robot-assisted prostatectomy: experience in one institution

    BJU Int

    (2003)
  • T.R. Bandaragoda et al.

    Text mining for personalised knowledge extraction from online support groups

    J Assoc Inf Sci Technol

    (2018 (In print))
  • T. Bandaragoda et al.

    The Patient-Reported Information Multidimensional Exploration (PRIME) framework for investigating emotions and other factors of prostate cancer patients with low intermediate risk based on online cancer support group discussions

    Ann Surg Oncol

    (2018)
  • W. Ranasinghe et al.

    A novel framework for automated, intelligent extraction and analysis of online support group discussions for cancer related outcomes

    BJU Int

    (2017)
  • A. Jaulim et al.

    A comparison of operative and margin outcomes from surgeon learning curves in robot assisted radical prostatectomy in a changing referral practice

    Ann R Coll Surg Engl

    (2018)

    • Cited by (12)

    • Empathic conversational agents for real-time monitoring and co-facilitation of patient-centered healthcare

      2022, Future Generation Computer Systems
      Citation Excerpt :

      The conversations are rich with patient emotions, their concerns and thoughts which are crucial to elevating the current PCC applications. The emotion expressed in online groups can be used to investigate different patient behavior patterns [20] and even use for clinical studies to assess the perception of different treatment types [21,22]. As the customization and individualization of PCC come from the patient-reported information, it is essential to investigate the emotions and concerns expressed in these conversations.

    • Churn Prediction of Clinical Decision Support Recommender System

      2023, Smart Innovation, Systems and Technologies

    • Artificial intelligence and machine learning in prostate cancer patient management—current trends and future perspectives

      2021, Diagnostics

    • Beyond social media analytics: Understanding human behaviour and deep emotion using self structuring incremental machine learning

      2020, arXiv

    • The application of artificial intelligence in prostate cancer management-What improvements can be expected? A systematic review

      2020, Applied Sciences (Switzerland)

    • Myth busting patient's pain: Comparing robotic-assisted verses open radical prostatectomies

      2020, Gland Surgery
    View all citing articles on Scopus
    View full text
    © 2018 Elsevier Inc. All rights reserved.