User:Sutrux2/sandbox

Working Intro:
With the ever increasing integration and accuracy of Artificial Intelligence (AI) as a diagnostic and analytical tool in the world of medicine, it has become possible to tackle previously insurmountable barriers.

First contact for many patients has become the Emergency Department (ED) as primary care physicians have become increasingly difficult to access for many patients. With the added pressure of increased ED crowding over the last decade and ED staff becoming hard-pressed for time and resources, having a system in place to prevent mistakes is crucial. One challenge physicians face, considering such high volumes and few resources, is to accurately triage patients to be either admitted or discharged. It is also necessary to overcome physician bias and human error, to which no effective solutions currently exist. (<-- I'm not sure yet about the certainty of this sentence)

In this study, we examined over 500,000 patients discharged from the ED over the span of 6 years. Despite best efforts, we found 600, or 1/1000 patients to have died within only 7 days after discharge. However, using the data and with the help of a state of the art algorithm developed in 2017, we built a prediction tool as a proof of concept that can alert clinicians if a patient is at risk of having a life threatening event after discharge. (I have to expand somehow, any ideas on more talking points here?)

2011 2006 1994

Notes:
(AI is awesome, and its being used more in medicine to overcome stuff we couldn't do before

Within medicine, the ED is has been identified to be increasingly first line to patients, and while docs are excellent, bias and human error have to be a part of the equation.

One of the challenges for physicians, esp since there is so much stress and such high volumes, is to accurately triage every patient either to be admitted or discharged. And despite best efforts, about 1/1000 patients on average die within 7 days of discharge. Currently, there is no system in place to catch such errors.

In this study, we have developed a prediction tool that spits out a score of how likely a pt is to die out of hospital. )

The Emergency Department (ED) is arguably one of the busiest and most fast-paced environments for medical professionals.

However, the two primary outcomes of the ED are quite straightforward, every visit ends in either admission or discharge.

With the increase in ED crowding over the last decade , often due to a decreasing access to primary care physicians , ED staff become increasingly hard-pressed for time and resources.

''With the increasing stress on ED staff and Emergency Medicine Physicians (EMPs) to filter as many patients as possible, there is more room for mistakes to occur. And patients that should be hospitalized can sometimes be inadvertently discharged from the ED.''

After reviewing over 500,000 patient discharges from the ED over the span of 6 years, about 600 were found to have died within only 7 days of discharge.

In this study, we reviewed over 500,000 individuals and found patients that were discharged and died at home, both within 48 hours (174 pts), within 7 days (600 pts), and within 30 days (# pts) and assessing what could have led physicians to make these mistakes, and what possible steps we could take to address and correct them in the future.""500,000 pt that were discharged

about 600 (1/1000) died within 7 days

algorithm has 95% sensitivity, 85% specificity

Sensitivity - out of those that died, how many did you predict will die

Specificity - out of those that didn’t die, how many wouldn’t you say will (PPV correlate)

PPV is 5% (don’t talk about it)

Will be alarmed for 20 pts where 1 who will die. (be more positive)

Make sure to read about if there were any studies done about this or like this.

Methods:
KLANG

Background diagnosis with dementia, no end of life diseases.

Table 1: Demographics: sex, age, background diagnoses (breakdown by %).

Table 2: CC, top 5, # pts that had same CC, % of pts that discharged at home.

Table 3: Vital signs: HTN, hypoxemia, fever, BP, pulse, disposition, pain, ESI (read about and describe about it)

Table 4: Labs

Results:
KLANG

Big ER, 5 year retrospectively investigated digital records of all pts in tertiary hospital (neurosurgery facility) from Jan 1 2013 - Dec 31 2017

Looked at variables: age, sex, etc.

Finding what's interesting - something will be interesting.

Discussion:
Each line you write in methods, you write in the results. Here only touch interesting points from those sections. Do not say that they are end of life, but rather say that they are oncological stage 4 etc.