Performance and clinical applicability of machine learning in liver computed tomography imaging: a systematic review.

OBJECTIVES: Machine learning (ML) for medical imaging is emerging for several organs and image modalities. Our objectives were to provide clinicians with an overview of this field by answering the following questions: (1) How is ML applied in liver computed tomography (CT) imaging? (2) How well do ML systems perform in liver CT imaging? (3) What are the clinical applications of ML in liver CT imaging? METHODS: A systematic review was carried out according to the guidelines from the PRISMA-P statement. The search string focused on studies containing content relating to artificial intelligence, liver, and computed tomography. RESULTS: One hundred ninety-one studies were included in the study. ML was applied to CT liver imaging by image analysis without clinicians' intervention in majority of studies while in newer studies the fusion of ML method with clinical intervention have been identified. Several were documented to perform very accurately on reliable but small data. Most models identified were deep learning-based, mainly using convolutional neural networks. Potentially many clinical applications of ML to CT liver imaging have been identified through our review including liver and its lesion segmentation and classification, segmentation of vascular structure inside the liver, fibrosis and cirrhosis staging, metastasis prediction, and evaluation of chemotherapy. CONCLUSION: Several studies attempted to provide transparent result of the model. To make the model convenient for a clinical application, prospective clinical validation studies are in urgent call. Computer scientists and engineers should seek to cooperate with health professionals to ensure this. KEY POINTS: • ML shows great potential for CT liver image tasks such as pixel-wise segmentation and classification of liver and liver lesions, fibrosis staging, metastasis prediction, and retrieval of relevant liver lesions from similar cases of other patients. • Despite presenting the result is not standardized, many studies have attempted to provide transparent results to interpret the machine learning method performance in the literature. • Prospective studies are in urgent call for clinical validation of ML method, preferably carried out by cooperation between clinicians and computer scientists.

Group-based multimodal physical therapy in women with chronic pelvic pain: A randomized controlled trial.

INTRODUCTION: Chronic pelvic pain in women is a complex condition, and physical therapy is recommended as part of a broader treatment approach. The objective of this study was to compare structured group-based multimodal physical therapy in a hospital setting (intervention group) with primary-care physical therapy (comparator group) for women with chronic pelvic pain. MATERIAL AND METHODS: Women aged 20-65 years with pelvic pain ≥6 months and referred for physical therapy were eligible. The primary outcome measure was change in the mean pelvic pain intensity from baseline to 12 months, measured using the numeric rating scale (0-10). Secondary outcomes were changes in scores of "worst" and "least" pain intensity, health-related quality of life, movement patterns, pain-related fear of movements, anxiety and depression, subjective health complaints, sexual function, incontinence, and obstructed defecation. The differences between the groups regarding change in scores were analyzed using the independent t test and Mann-Whitney U test. Sensitivity analysis of the primary outcome was performed with a linear regression model adjusted for the baseline value. A P value <.05 was considered statistically significant. RESULTS: Of the 62 women included, 26 in the intervention group and 25 in the comparator group were available after 12 months for data collection and analysis. The difference between the groups for change in the mean pain intensity score was -1.2 (95% CI -2.3 to -0.2; P = .027), favoring the intervention group. The intervention group showed greater improvements in respiratory patterns (mean difference 0.9; 95% CI 0.2-1.6; P = .015) and pain-related fear of movements (mean difference 2.9; 95% CI -5.5 to -0.3; P = .032), and no significant differences were observed between the groups for the other secondary outcomes. CONCLUSIONS: Although the reduction in the mean pelvic pain intensity with group-based multimodal physical therapy was significantly more than with primary-care physical therapy, the difference in the change between the groups was less than expected and the clinical relevance is uncertain.

Predicting colorectal surgical complications using heterogeneous clinical data and kernel methods.

OBJECTIVE: In this work, we have developed a learning system capable of exploiting information conveyed by longitudinal Electronic Health Records (EHRs) for the prediction of a common postoperative complication, Anastomosis Leakage (AL), in a data-driven way and by fusing temporal population data from different and heterogeneous sources in the EHRs. MATERIAL AND METHODS: We used linear and non-linear kernel methods individually for each data source, and leveraging the powerful multiple kernels for their effective combination. To validate the system, we used data from the EHR of the gastrointestinal department at a university hospital. RESULTS: We first investigated the early prediction performance from each data source separately, by computing Area Under the Curve values for processed free text (0.83), blood tests (0.74), and vital signs (0.65), respectively. When exploiting the heterogeneous data sources combined using the composite kernel framework, the prediction capabilities increased considerably (0.92). Finally, posterior probabilities were evaluated for risk assessment of patients as an aid for clinicians to raise alertness at an early stage, in order to act promptly for avoiding AL complications. DISCUSSION: Machine-learning statistical model from EHR data can be useful to predict surgical complications. The combination of EHR extracted free text, blood samples values, and patient vital signs, improves the model performance. These results can be used as a framework for preoperative clinical decision support.

Nationwide improvement of rectal cancer treatment outcomes in Norway, 1993-2010.

BACKGROUND: The Norwegian Rectal Cancer Project was initated in 1993 with the aims of improving surgery, decreasing local recurrence rates, improving survival, and establishing a national rectal cancer registry. Here we present results from the Norwegian Colorectal Cancer Registry (NCCR) from 1993 to 2010. MATERIAL AND METHODS: A total of 15 193 patients were diagnosed with rectal cancer in Norway 1993-2010, and were registered with clinical data regarding diagnosis, treatment, locoregional recurrences and distant metastases. Of these, 10 796 with non-metastatic disease underwent tumour resection. The results were stratified into five time periods, and the treatment outcomes were compared. Recurrence rates are presented for the 9785 patients who underwent curative major resection (R0/R1). RESULTS: Among all 15 193 patients, relative five-year survival increased from 54.1% in 1993-1997 to 63.4% in 2007-2010 (p < 0.001). Among the 10 796 patients with stage I-III disease who underwent tumour resection, from 1993-1997 to 2007-2010, relative five-year survival improved from 71.2% to 80.6% (p < 0.001). An increasing proportion of these patients underwent surgery at large-volume hospitals; and 30- and 100-day mortality rates, respectively, decreased from 3.0% to 1.4% (p < 0.001) and from 5.1% to 3.0% (p < 0.011). Use of preoperative chemoradiotherapy increased from 6.5% in 1993 to 39.0% in 2010 (p < 0.001). Estimated local recurrence rate after major resection (R0/R1) decreased from 14.5% in 1993-1997 to 5.0% in 2007-2009 (p < 0.001), and distant recurrence rate decreased from 26.0% to 20.2% (p < 0.001). CONCLUSION: Long-term outcomes from a national population-based rectal cancer registry are presented. Improvements in rectal cancer treatment have led to decreased recurrence rates of 5% and increased survival on a national level.

A prospective analysis of false positive events in a National Colon Cancer Surveillance Program.

BACKGROUND: The survival benefits of colon cancer surveillance programs are well delineated, but less is known about the magnitude of false positive testing. The objective of this study was to estimate the false positive rate and positive predictive value of testing as part of a surveillance program based on national guidelines, and to estimate the degree of testing and resource use needed to identify a curable recurrence. METHODS: Analysis of clinically significant events leading to suspicion of cancer recurrence, false positive events, true cancer recurrences, time to confirmation of diagnosis, and resource use (radiology, blood samples, colonoscopies, consultations) among patients included in a randomised colon cancer surveillance trial. RESULTS: 110 patients surgically treated for colon cancer were followed according to national guidelines for 1884 surveillance months. 1105 tests (503 blood samples, 278 chest x-rays, 209 liver ultrasounds, 115 colonoscopies) and 1186 health care consultations were performed. Of the 48 events leading to suspicion of cancer recurrence, 34 (71%) represented false positives. Thirty-one (65%) were initiated by new symptoms, and 17 (35%) were initiated by test results. Fourteen patients had true cancer recurrence; 7 resections of recurrent disease were performed, 4 of which were successful R0 metastasis Resections. 276 tests and 296 healthcare consultations were needed per R0 resection; the cost per R0 surgery was £ 103207. There was a 29% probability (positive predictive value) of recurrent cancer when a diagnostic work-up was initiated based on surveillance testing or patient complaints. CONCLUSION: We observed a high false positive rate and low positive predictive value for significant clinical events suggestive of possible colorectal cancer relapse in the setting of a post-treatment surveillance program based on national guidelines. Providers and their patients should have an appreciation for the modest positive predictive value inherent in colorectal cancer surveillance programs in order to make informed choices, which maximize quality of life during survivorship. Better means of tailoring surveillance programs based on patient risk would likely lead to more effective and cost-effective post-treatment follow-up. TRIAL REGISTRATION: ClinicalTrials.gov identifier NCT00572143. Date of trial registration: 11th of December 2007.

Surgical telementoring in knowledge translation--clinical outcomes and educational benefits: a comprehensive review.

BACKGROUND: Surgical telementoring has been reported for decades. However, there exists limited evidence of clinical outcome and educational benefits. OBJECTIVE: To perform a comprehensive review of surgical telementoring surveys published in the past 2 decades. RESULTS: Of 624 primary identified articles, 34 articles were reviewed. A total of 433 surgical procedures were performed by 180 surgeons. Most common telementored procedures were laparoscopic cholecystectomy (57 cases, 13%), endovascular treatment of aortic aneurysm (48 cases, 11%), laparoscopic colectomy (32 cases, 7%), and nefrectomies (41 cases, 9%). In all, 167 (38%) cases had a laparoscopic approach, and 8 cases (5%) were converted to open surgery. Overall, 20 complications (5%) were reported (liver bleeding, trocar port bleeding, bile collection, postoperative ileus, wound infection, serosa tears, iliac artery rupture, conversion open surgery). Eight surveys (23%) have structured assessment of educational outcomes. Telementoring was combined with simulators (n = 2) and robotics (n = 3). Twelve surveys (35%) were intercontinental. Technology satisfaction was high among 83% of surgeons. CONCLUSION: Few surveys have a structured assessment of educational outcome. Telementoring has improved impact on surgical education. Reported complication rate was 5%.

Sacral nerve stimulation.

BACKGROUND: Sacral nerve stimulation implies electrical stimulation of a sacral nerve root by an electrode and a pacemaker. Within the past few years, sacral nerve stimulation has become a possible treatment option for selected patients with urinary retention, urinary incontinence, anal incontinence and constipation. The method is furthermore being tested for several other conditions. MATERIAL AND METHODS: The article presents the method and treatment results following various indications based on the authors' own experience and non-systematic PubMed search. RESULTS: During a test period an external pacemaker is used for 3-30 days, with length of test differing according to the indication. A positive test (improvement of symptoms by 50 % or more) is achieved by 70-90 % of patients with anal incontinence, 70 % with urinary non-obstructive retention, 52-77 % with urinary urge incontinence and 43-72 % with constipation. Sacral nerve stimulation may also be effective in patients with chronic pelvic pain. Following implantation of a pacemaker a sustainable effect is seen in 50-90 % of patients with a positive test. Up to 75 % of patients will need repeated follow-up including pacemaker reprogramming or reoperations due to diminished effect. The longevity of the pacemaker is 3-10 years, and it must be replaced operatively when the battery has depleted. INTERPRETATION: Treatment with sacral nerve stimulation may be efficient over time in patients with various pelvic floor dysfunctions, especially anal incontinence and non-obstructive urinary retention. Most of the patients will need close follow-up in order to maintain an optimal result.

International preoperative rectal cancer management: staging, neoadjuvant treatment, and impact of multidisciplinary teams.

BACKGROUND: Little is known regarding variations in preoperative treatment and practice for rectal cancer (RC) on an international level, yet practice variation may result in differences in recurrence and survival rates. METHODS: One hundred seventy-three international colorectal centers were invited to participate in a survey of preoperative management of rectal cancer. RESULTS: One hundred twenty-three (71%) responded, with a majority of respondents from North America, Europe, and Asia. Ninety-three percent have more than 5 years' experience with rectal cancer surgery. Fifty-five percent use CT scan, 35% MRI, 29% ERUS, 12% digital rectal examination and 1% PET scan in all RC cases. Seventy-four percent consider threatened circumferential margin (CRM) an indication for neoadjuvant treatment. Ninety-two percent prefer 5-FU-based long-course neoadjuvant chemoradiation therapy (CRT). A significant difference in practice exists between the US and non-US surgeons: poor histological differentiation as an indication for CRT (25% vs. 7.0%, p = 0.008), CRT for stage II and III rectal cancer (92% vs. 43%, p = 0.0001), MRI for all RC patients (20% vs. 42%, p = 0.03), and ERUS for all RC patients (43% vs. 21%, p = 0.01). Multidisciplinary team meetings significantly influence decisions for MRI (RR = 3.62), neoadjuvant treatment (threatened CRM, RR = 5.67, stage II + III RR = 2.98), quality of pathology report (RR = 4.85), and sphincter-saving surgery (RR = 3.81). CONCLUSIONS: There was little consensus on staging, neoadjuvant treatment, and preoperative management of rectal cancer. Regular multidisciplinary team meetings influence decisions about neoadjuvant treatment and staging methods.

[Overactive pelvic floor syndrome]. / Overaktiv bekkenbunn-syndrom.

BACKGROUND: The etiology of the overactive pelvic floor syndrome is not fully understood and no gold standards are available for diagnosis or treatment. The article presents an overview of literature, and discusses diagnostics and treatment. MATERIAL AND METHODS: Literature was identified through a non-systematic search in PubMed, and discussed in light of the authors' clinical experience with the patient group. RESULTS: The main symptoms of overactive pelvic floor syndrome are pain and defecation difficulties; the latter often leads to chronic constipation. Other symptoms depend on which parts of the pelvic floor that are most affected. Pain is often chronic and ranges from mild to severe; it is aggravated by micturition, sexual intercourse, orgasm, defecation and sitting on hard surfaces, and reduces the ability to work and quality of life in general. Injection of Botulinum toxin in the pelvic floor muscles seems to alleviate pain in many patients. Physiotherapy of the pelvic floor and treatment offered by pain clinics can also be useful. INTERPRETATION: A close cooperation between gastroenterologists, surgeons, urologists, gynecologists, neurologists, physiotherapists and possibly pain clinics is important to improve the situation for these patients.

Overcoming distance: video-conferencing as a clinical and educational tool among surgeons.

BACKGROUND: Since the 1960s, there has been substantial development in the uses of video-conferencing (VC) among medical personnel, including surgeons who have adopted the technology. METHODS: A report on our own experience with VC was combined with a comprehensive PubMed search with the key words telepresence, video-conferencing, video-teleconferencing, telementoring and surgery, trauma, follow-up, education, and multidisciplinary teams. A search through two peer-reviewed telemedicine journals -- Journal of Telemedicine and Telecare and Telemedicine and e-Health Journal -- and references of all included papers and identified additional reports was conducted. RESULTS: A total of 517 articles were identified with 51 relevant manuscripts, which included the key phrases. VC is widely used among surgeons for telementoring surgical procedures and in trauma and emergency medicine. Furthermore, VC is widely used by multidisciplinary teams and for the follow-up of patients after surgery. CONCLUSIONS: VC is a common clinical tool for surgeons and provides a great opportunity to alter surgical practice and to offer patients the best expertise in surgical treatment despite long distances, especially in rural areas.

Standardized postoperative pathway: accelerating recovery after ileostomy closure.

PURPOSE: In this study we evaluated the outcome of a standardized enhanced recovery program in patients undergoing ileostomy closure. METHODS: Forty-two patients underwent ileostomy closure by a single surgeon and were managed by a standardized postoperative care pathway. On the first postoperative day, patients received oral analgesia and a soft diet. Discharge was based on standard criteria previously published for laparoscopic colectomy patients. Results were recorded prospectively in an Institutional Review Board-approved database, including demographics, operative time, blood loss, complications, length of stay, and readmission data. RESULTS: The median operative time and blood loss were 60 minutes and 17.5 mL, respectively, and median hospital stay was 2 days. Twenty-nine patients (69 percent) were discharged by postoperative Day 2. The complication rate was 23.8 percent; complications included prolonged postoperative ileus (n = 3), early postoperative small-bowel obstruction (n = 1), mortality not related to ileostomy closure (n = 1), minor bleeding (n = 1), wound infection (n = 1), incisional hernia (n = 1), diarrhea (n = 1), dehydration (n = 1). The 30-day readmission rate was 9.5 percent (n = 4). Two patients had reoperation within 30 days for small-bowel obstruction and a wound infection. CONCLUSIONS: Ileostomy closure patients managed with postoperative care pathways can have a short hospital stay with acceptable morbidity and readmission rates.

Surgical treatment for rectal cancer: an international perspective on what the medical gastroenterologist needs to know.

Rectal cancer accounts for one third of all colorectal cancers. The age adjusted death rates from colorectal cancer have declined over recent decades due to a combination of colorectal cancer screening, improved diagnostic tests, improved standardized surgical technique, improved medical support, neoadjuvant chemotherapies and radiation treatment or combinations of these. Because of complex treatment algorithms, use of multidisciplinary teams in the management of rectal cancer patients has also been popularized. Medical gastroenterologists performing colonoscopies are frequently the first health care provider to raise the suspicion of a rectal cancer. Although the diagnosis depends on histological confirmation, the endoscopic presentation is almost diagnostic in many cases. In order to meet the patient's immediate needs for information, it is important that the endoscopist has knowledge about the investigations and treatment options that will be required for their patient. The aim of this paper is to describe the modern preoperative investigations and operative procedures commonly offered to rectal cancer patients taking into account perspectives of three colorectal surgeons, practicing in the USA, Europe and Asia.

Accuracy of Helicobacter pylori serology in two peptic ulcer populations and in healthy controls.

AIM: To estimate the test characteristics of Helicobacter pylori (H pylori) serology and of C14-urea breath test (C14-UBT) in two different peptic ulcer populations and in community controls. Second, the aim was to explore the association between the level of H pylori IgG antibodies and severity of inflammation as to active peptic ulceration in the same populations. METHODS: Vagotomized (n = 83), medically treated peptic ulcer patients (n = 73) and one reference group of community controls (n = 88) were gastroscoped. H pylori status was determined by histology, bacterial growth, C14-UBT and serology. Based on the updated Sydney System, cumulative scores from biopsies from the prepyloruos, incisura angularis, corpus and fundus were calculated. RESULTS: The prevalence of H pylori infection varied from 70% to 79%. The C14-UBT had high accuracy compared to the serology test. The sensitivity of the serology test was good, but the specificity was low (41%-71%). The association between H pylori IgG antibodies and scores of gastric mucosal inflammation and current or previous peptic ulcer were weak. CONCLUSION: The accuracy of C14-UBT to diagnose H pylori infection was good, and the clinical utility of a negative H pylori serology test was substantial, while the gain in clinical information of a positive test was meagre. Positive H pylori titres could not distinguish between subjects with or those without active peptic ulceration.

Should the surgeon or the general practitioner (GP) follow up patients after surgery for colon cancer? A randomized controlled trial protocol focusing on quality of life, cost-effectiveness and serious clinical events.

BACKGROUND: All patients who undergo surgery for colon cancer are followed up according to the guidelines of the Norwegian Gastrointestinal Cancer Group (NGICG). These guidelines state that the aims of follow-up after surgery are to perform quality assessment, provide support and improve survival. In Norway, most of these patients are followed up in a hospital setting. We describe a multi-centre randomized controlled trial to test whether these patients can be followed up by their general practitioner (GP) without altering quality of life, cost effectiveness and/or the incidence of serious clinical events. METHODS AND DESIGN: Patients undergoing surgery for colon cancer with histological grade Dukes's Stage A, B or C and below 75 years of age are eligible for inclusion. They will be randomized after surgery to follow-up at the surgical outpatient clinic (control group) or follow-up by the district GP (intervention group). Both study arms comply with the national NGICG guidelines. The primary endpoints will be quality of life (QoL) (measured by the EORTC QLQ C-30 and the EQ-5D instruments), serious clinical events (SCEs), and costs. The follow-up period will be two years after surgery, and quality of life will be measured every three months. SCEs and costs will be estimated prospectively. The sample size was 170 patients. DISCUSSION: There is an ongoing debate on the best method of follow-up for patients with CRC. Due to a wide range of follow-up programmes and paucity of randomized trials, it is impossible to draw conclusions about the best combination and frequency of clinic (or family practice) visits, blood tests, endoscopic procedures and radiological examinations that maximize the clinical outcome, quality of life and costs. Most studies on follow-up of CRC patients have been performed in a hospital outpatient setting. We hypothesize that postoperative follow-up of colon cancer patients (according to national guidelines) by GPs will not have any impact on patients' quality of life. Furthermore, we hypothesize that there will be no increase in SCEs and that the incremental cost-effectiveness ratio will improve. TRIAL REGISTRATION: This trial has been registered at ClinicalTrials.gov. The trial registration number is: NCT00572143.

The one-stop trial: does electronic referral and booking by the general practitioner (GPs) to outpatient day case surgery reduce waiting time and costs? A randomized controlled trial protocol.

BACKGROUND: Waiting time and costs from referral to day case outpatient surgery are at an unacceptably high level. The waiting time in Norway averages 240 days for common surgical conditions. Furthermore, in North Norway the population is scattered throughout a large geographic area, making the cost of travel to a specialist examination before surgery considerable. Electronic standardised referrals and booking of day case outpatient surgery by GPs are possible through the National Health Network, which links all health care providers in an electronic network. New ways of using this network might reduce the waiting time and cost of outpatient day case surgery. MATERIALS AND METHODS: In a randomised controlled trial, selected patients (inguinal hernia, gallstone disease and pilonidal sinus) referred to the university hospital are either randomised to direct electronic referral and booking for outpatient surgery (one stop), or to the traditional patient pathway where all patients are seen at the outpatient clinic several weeks ahead of surgery. Consultants in gastrointestinal surgery designed standardised referral forms and guidelines. New software has been designed making it possible to implement referral forms, guidelines and patient information in the GP's electronic health record. For "one-stop" referral, GPs must provide mandatory information about the specific condition. Referrals were linked to a booking system, enabling the GPs to book the hospital, day and time for outpatient surgery. The primary endpoints are waiting time and costs. The sample size calculation was based on waiting time. A reduction in waiting time of 60 days (effect size), 25%, is significant, resulting in a sample size of 120 patients in total. DISCUSSION: Poor communication between primary and secondary care often results in inefficiencies and unsatisfactory outcomes. We hypothesised that standardised referrals would improve the quality of information, making it feasible to use a one-stop approach for all patients undergoing surgery on an outpatient basis for inguinal hernia, pilonidal sinus and gallstones. In this study we wanted to investigate the waiting time and cost-effectiveness of direct electronic referral and booking of outpatient surgery compared to the traditional patient pathway, where the patient is seen at the outpatient clinic prior to surgery. TRIAL REGISTRATION: This trial has been registered at ClinicalTrials.gov. The trial registration number is: NCT00692497.

[Assessment of anal incontinence in adults]. / Utredning av anal inkontinens hos voksne.

BACKGROUND: Anal incontinence is involuntary loss of stool and flatus through the anal canal. The condition can be severely debilitating to those affected. MATERIAL AND METHODS: The article is based on relevant literature and the authors' clinical experience. RESULTS: The first examination should include an anorectoscopy as part of a broader clinical examination, and the aim should be to find the cause and describe the condition. An assessment in hospital should comprise endoanal ultrasound for imaging and grading of the incontinence severity with a score system. CONCLUSION: Simple preliminary investigations and eventually more specialized investigations in a specialist clinic give a good impression of the condition, and is the basis for choosing treatment and later evaluation.

[Anal incontinence--assessment of a new treatment]. / Anal inkontinens--utprøvning av en ny behandlingsmetode.

BACKGROUND: Anal incontinence affects 1.4 % of the general population. Anal injections is a new treatment option for this condition; different techniques and bulking agents are used. The article presents a pilot-study with a new bulking agent (hyaluronic acid and dextranomere) and a new injection technique. MATERIAL AND METHODS: An anascope was used to inject Zuidex (4 x 1. 4 mL) submucosally (proximal to the dentate line and distally to the puborectal muscle) in 4 patients with severe faecal incontinence, who were deemed unfit for other treatment. No anaesthesia or antibiotic-prophylaxis was used. RESULTS: All patients tolerated the treatment well, and there were no adverse events. The treatment had an effect in 3 of 4 patients; there was a median fall in St. Mark's score of 3. 5 points. INTERPRETATION: The injection technique was well tolerated, easy to perform within an outpatient setting and with promising short-term results. The method has been implemented in a randomized, controlled trial.

Using anchors from free text in electronic health records to diagnose postoperative delirium.

OBJECTIVES: Postoperative delirium is a common complication after major surgery among the elderly. Despite its potentially serious consequences, the complication often goes undetected and undiagnosed. In order to provide diagnosis support one could potentially exploit the information hidden in free text documents from electronic health records using data-driven clinical decision support tools. However, these tools depend on labeled training data and can be both time consuming and expensive to create. METHODS: The recent learning with anchors framework resolves this problem by transforming key observations (anchors) into labels. This is a promising framework, but it is heavily reliant on clinicians knowledge for specifying good anchor choices in order to perform well. In this paper we propose a novel method for specifying anchors from free text documents, following an exploratory data analysis approach based on clustering and data visualization techniques. We investigate the use of the new framework as a way to detect postoperative delirium. RESULTS: By applying the proposed method to medical data gathered from a Norwegian university hospital, we increase the area under the precision-recall curve from 0.51 to 0.96 compared to baselines. CONCLUSIONS: The proposed approach can be used as a framework for clinical decision support for postoperative delirium.

Analysis of free text in electronic health records for identification of cancer patient trajectories.

With an aging patient population and increasing complexity in patient disease trajectories, physicians are often met with complex patient histories from which clinical decisions must be made. Due to the increasing rate of adverse events and hospitals facing financial penalties for readmission, there has never been a greater need to enforce evidence-led medical decision-making using available health care data. In the present work, we studied a cohort of 7,741 patients, of whom 4,080 were diagnosed with cancer, surgically treated at a University Hospital in the years 2004-2012. We have developed a methodology that allows disease trajectories of the cancer patients to be estimated from free text in electronic health records (EHRs). By using these disease trajectories, we predict 80% of patient events ahead in time. By control of confounders from 8326 quantified events, we identified 557 events that constitute high subsequent risks (risk > 20%), including six events for cancer and seven events for metastasis. We believe that the presented methodology and findings could be used to improve clinical decision support and personalize trajectories, thereby decreasing adverse events and optimizing cancer treatment.