12 lines
1.8 KiB
Plaintext
12 lines
1.8 KiB
Plaintext
08:34 DocFriend1 To answer your question, generally, yes, models like this have great clinical implication.
|
||
08:35 DocFriend1 You’re building something that’s akin to an early warning system model, like EPIC early warning systems or NEWS in the UK.
|
||
08:37 DocFriend1 Early warning systems allow physicians and care practitioners to proactively intervene on patients before life threatening events. This is especially useful because clinical deterioration is an actual problem in most care settings and is a significant contributor to morbidity and mortality
|
||
08:38 DocFriend1 Funny enough, my original thesis idea was evaluating the effectiveness of early warning systems
|
||
08:38 DocFriend1 In real world clinical settings
|
||
08:39 DocFriend1 I was going to use a regression discontinuity design.
|
||
08:40 DocFriend1 One of the biggest issues with this, is alert fatigue, most physicians are becoming increasingly burnout by constant alerts.
|
||
08:41 DocFriend1 In comparison to other models, off the top of my head, your sepsis tracking through the SIRS criteria is good, most models give the vitals and labs, so physicians would have to evaluate that by themselves and most forget or don’t think about it
|
||
08:42 DocFriend1 However, if your models does it autonomously and provides a report, that’s step in the right direction, I can do more research into models and see how they compare.
|
||
08:44 DocFriend1 Another thing to note is that EWS use a scoring method and gives an acuity score, once a patient scores above a certain number, the alert is triggered.
|
||
08:45 DocFriend1 The score is calculated from 5 or 6 components depending on the system (company) vitals mostly.
|
||
08:45 DocFriend1 Yours seemingly would track each sign and compare it against a clinical standard or range, and if that exceeds, the alert is given. |