Almost Election Day

Not that I'm going to write anything about it.  I'm just trying to vary the titles of the posts from just simply being variable descriptions of how reading is going.  Of course, that is exactly what the content is going to be.

I've made the rounds back to the "Flesch Cards", only we've simply been doing them on the computer.  I zoom in on the text files so that only one full word is seen on the screen and I'll either tape a piece of paper over half the screen or cover it with a book so that he can't see the picture.  I went through all the old ones that we had previously printed out and he got all of them right on the first try (I think that is 1 through 3 or 4).  We are up to number 8 and he can read ~90% of the words on the first try without any hesitation.  What's nice is that if he isn't sure he will default to sounding out the letters which I like to see him use rather than just guessing on the word based on context and the first and last letters (which he will often do).

As far as independently reading things on his own, I would say he is still on a GRL C or D level, where these books are completely readable by him alone on the first try.  Sometimes I do have to remind him that the other page has words too as he often seems more interested in the pictures.  I've been checking out a lot of the "Holiday" publishing company's children books since they are pretty consistent.

I have definitely noticed him reading things like signs or maps in the real world more often, so hopefully his interested in self-guided reading will pick up (which is what I would really like, as it would allow me to sort of just pop in when he has trouble with a word).  I'm excited to get to the concept of two syllable words because I think it will change his method a lot when he realizes how modular English can be (emphasis on the can).

To remind myself in the future for benchmarking with our other children:  a couple weeks ago I revisited a youtube video of frequent sight words and he knew all of them without hesitation, but I can't remember exactly when that was.  It may have even been before the last post.

In programming news I thought it might be fun to make a little python program to help diagnose impending death in actively dying patients based on research of David Hui and his group at MD Anderson (see the Cancer and Oncologist papers for their signs and symptoms).  It's been a nice little review in applying likelihood ratios to pre-test probabilities and how to calculate them.  The huge problem I keep running into is how to accurately assess a pretest probability.  For one thing it can have a larger affect than any positive or negative LR.  Secondly, if it seems that by objectively finding a good pretest probability, I'm already most of the way there to answering my question in the first place!  Otherwise, what does it really mean when I'm saying a patient has a 30% vs a 70% chance of dying in the next 3 days?  Both are quite "large" numbers, yet after a positive LR of 10.0, the former post test prob is 81% and the latter 95%.  How useful is telling a family that their loved one has an 81% vs 95% chance of dying in the next 3 days?  I guess that might actually show some usefulness because no matter what it shows that a positive sign means death is likely.  On the other hand, let's take 1% pretest vs 10% pretest.  The absolute difference in these numbers is small but the resultant post test probs (9% vs 51%) actually tell very different stories.  How confident can I be that a patient has a 1% chance of dying vs a 10%?  This is where I am stuck and looking for answers in the research.