As I read this abstract, my mind went into a whole different area: machine learning. Technology has come so far in the last 5 years that at some point, we need to discuss the idea of machine learning to help improve the care we provide.
More and more research is being dedicated to predictive factors. I can appreciate the need to be able to predict outcomes. The one thing with FOTO that I truly love is the predictive ability based on the risk adjustment process. At the same time, my mind wonders.
As we know, having the ability to be able to predict an outcome AND the best path of care would have a high amount of value. In the near future, we will begin to see more money and time spent in designing systems to include machine learning. I wanted to take time to share machine learning and how it could be helpful in the medical world.
After listening to Suchi Saria speak, it really seemed to me that systematic reviews may not be the best option when it comes to physical prognostic factors. What would it look like if we began to implement machine learning? We’d have a ton of data and information points within the electronic medical record. We’d capture the real world, all encompassing, messy information versus the few data points and factors we seem to believe have the most value. I feel excited about what could be learned and how quickly we could learn based on real-life data.
You’ll find the abstract to the recent study below.
Physical prognostic factors predicting outcome following lumbar discectomy surgery: systematic review and narrative synthesis.
Success rates for lumbar discectomy are estimated as 78-95% patients at 1-2 years post-surgery, supporting its effectiveness. However, ongoing pain and disability is an issue for some patients, and recurrence contributing to reoperation is reported. It is important to identify prognostic factors predicting outcome to inform decision-making for surgery and rehabilitation following surgery. The objective was to determine whether pre-operative physical factors are associated with post-operative outcomes in adult patients [≥16 years old] undergoing lumbar discectomy or microdiscectomy.
A systematic review was conducted according to a registered protocol [PROSPERO CRD42015024168]. Key electronic databases were searched [PubMed, CINAHL, EMBASE, MEDLINE, PEDro and ZETOC] using pre-defined terms [e.g. radicular pain] to 31/3/2017; with additional searching of journals, reference lists and unpublished literature. Prospective cohort studies with ≥1-year follow-up, evaluating candidate physical prognostic factors [e.g. leg pain intensity and straight leg raise test], in adult patients undergoing lumbar discectomy/microdiscectomy were included. Two reviewers independently searched information sources, evaluated studies for inclusion, extracted data, and assessed risk of bias [QUIPS]. GRADE determined the overall quality of evidence.
1189 title and abstracts and 45 full texts were assessed, to include 6 studies; 1 low and 5 high risk of bias. Meta-analysis was not possible [risk of bias, clinical heterogeneity]. A narrative synthesis was performed. There is low level evidence that higher severity of pre-operative leg pain predicts better Core Outcome Measures Index at 12 months and better post-operative leg pain at 2 and 7 years. There is very low level evidence that a lower pre-operative EQ-5D predicts better EQ-5D at 2 years. Low level evidence supports duration of leg pain pre-operatively not being associated with outcome, and very low-quality evidence supports other factors [pre-operative ODI, duration back pain, severity back pain, ipsilateral SLR and forward bend] not being associated with outcome [range of outcome measures used].
An adequately powered low risk of bias prospective observational study is required to further investigate candidate physical prognostic factors owing to existing low/very-low level of evidence.