From Evelinn: The CO catchment area is concerned with HPV, so it should be present at some point on the site.

Split from #15

Say where sources came from. Defer to CancerInFocus's sources page rather than duplicating data.

Particularly the overlays/markers/locations multi-select might end up with a lot of options. Right now, it does have "type-ahead", where typing some characters while the box is focused jumps to the next option that starts with those characters. Operating systems like windows and linux do this too in their file explorer windows. But I think most people don't know about that functionality, and it also doesn't support searching for characters at any position in the text; it has to start with the characters.

Convert to this component instead:
https://headlessui.com/react/combobox

From Jake Walsh's email:

The first idea has to do with adding a county snapshot button to each county. We already have county snapshots that include information about some cancer incidence and mortality, but the issue is these snapshots have old data and need to be updated by hand. I want to see if it's possible to add a snapshot button directly to ECCO so that users can make their own county snapshots. The goal would be for the user to select which measures they want to look at for a specific county and then ECCO would automatically collect the data and visualize it. If that's too much, then I think just automatically visualizing some data about cancer incidence and mortality in the counties would be a good basepoint.

I think we'd realize this as a different page than the map, e.g. a page for a specific county that would show data from either all the measures or some user-selected subset. Since we currently just have the latest data for each measure, the data would likely be presented as a table. Perhaps once we start accumulating data over time (see #74, specifically the historical data question) we can explore other visualization options.

Rather than downloading a specific measure category/measure, allow the user to download the entire backing dataset in some to-be-decided format.

• wwc-15-jun-2021
• ccsp-15-jun-2021
• title-x-15-jun-2021
• chc-15-jun-2021
• rhc-15-jun-2021
• ihs-15-jun-2021
• genetic-coun-15-jun-2021
• all-treatment-15-jun-2021
• med-onc-15-jun-2021
• surg-onc-15-jun-2021
• rad-onc-15-jun-2021
• ped-onc-15-jun-2021
• trials-15-jun-2021
• all-support-services-15-jun-2021
• pt-rehab-15-jun-2021
• nutrition-15-jun-2021
• support-grp-15-jun-2021
• support-coun-15-jun-2021
• survivorship-15-jun-2021
• lodging-15-jun-2021
• financial-coun-15-jun-2021
• palliative-15-jun-2021
• hospice-15-jun-2021
• cancer-ed-29-jul-221
• accessories-aids-29-jul-21
• programs-events-29-jul-21
• fitness-29-jul-21
• home-family-care-29-jul-21
• transportation-29-jul-21
• pediatric-29-jul-21
• radiology

Also split this up into 3 to 5 (?) high level categories and multi-selects.

Hello,
Great dataset!
After selecting the environment category, you can select different measurements:
pct5G_35_3,
pct5G_7_1,
pctBB_1000_10,
pctBB_100_20.
What do they all mean?
Thank you very much in advance!
Best,
Etienne

I'm hoping that we can use the API not just for web development but also to access data. To that end, if you wanted to have an API endpoint that listed links to CSV files, that might be very user-friendly and also allow for a webpage that allows non-programmers to get access to underlying data.

cancerinfocus.org lets you select markers, regions, etc (let's call them "overlays) for things like lung cancer screening centers, toxic waste sites, food deserts, etc. These can be either exact points, or regions/areas (though they're generally smaller than countries/tracts).

Which of these sources of info do we want to add, and where would we get the data? I imagine this is something where we could add things to our hearts content based on what we find publicly available.

Sean had mentioned this link, but this seems to not provide what I'd need to display on a map. We'd need some kind of geo coordinates, along with any other info on a per point basis (e.g. cancerinfocus.org shows facility hours and phone number when you click on a cancer screening center marker).

About page, bottom of home page, etc.

Also any links in the footer?

@janolowery

Netlify throws a 404 if you access a subpage, e.g. "Sources": https://exploring-cancer-in-colorado.netlify.app/sources.

I assume this is because it hasn't been configured to send all requests to the root index where they can be handled by the router.

FWIW, I'm fine to close this as wontfix since it doesn't affect "prod" (e.g., https://ecco.cu-dbmi.dev/sources works) and it doesn't impact navigating around the site from the index, just coming in from a link with a subpath, but I thought I'd bring it up if there's any easy fix.

Currently, I manually run tools to scrape data from the various sources that contribute to the ECCO database.

We should have a pipeline set up that runs periodically and does the following:

1. pulls data from each source, if available
2. updates database, creates new db dump
3. loads new db dump into production

Possible Decisions:

• How often should the pipeline run? monthly?
• Should we include all data from each run in the database (e.g., "historical data"), or just the most recent data?
• If we do store historical data, do we make that accessible from the web UI?

Perhaps start as a google doc to keep list of where things come from and provide user-level and technical documentation.

Some of the cancer incidence and mortality measures are returning no data, and the legend as a result displays a percentage.

Example: https://ecco.cu-dbmi.dev/?level=county&category=cancerincidence&measure=Ovary&lat=38.98302&long=-106.16467&zoom=7.8

Images from Jan's report on the Teams channel (thanks, Jan):

So navigating directly to e.g. /about on deployed version of site properly redirects.

In the current implementation of the frontend and backend, the "Geographic level", "Measure category", and "Measure" dropdowns all automatically select the first available value, and there's no option to represent "no selection". Because of that, the "Download Data" button always retrieves data for a full selection of the level, category, and measure.

It's technically feasible and could be useful to return data for just the level and category, returning data across all the measures, as mentioned in #7 (comment). The schema of the downloaded CSV already has a "measure" column, which currently redundantly contains the currently-selected measure.

To support this new functionality, in the frontend we'd introduce a "no selection" option to the "Measure" dropdown. In the backend, if no measure was provided, we'd remove the code that filters to a particular measure and just return all the data for a category.

Currently the data for the locations is hardcoded as geojson into the frontend (src/api). It should ideally be moved to the backend, even if hardcoded there, for consistency.

For the sake of automating the ingest of this data later in #74, see #19 for how the data was initially scraped and added to the frontend.

Comboboxes probably not necessary in this case. In general, I feel they're mostly only necessary when you're searching a whole db of functionally "infinite" options, e.g. monarch phenotype explorer. If it's a finite set of options, and you wouldn't even know what to search for anyway until you see the options, it's less useful. It also reduces the clickable area for opening the dropdown, and I feel is probably less common of a UX pattern to users.

The "feedback" link is displayed when the page first opens, but many users (myself included) might click the "close box" on what appears to be an informational message.

It might be worth considering placing a "feedback" or "contact us" menu item next to the "about" menu item.

Doing so would result in one less click and no scrolling to get to a contact/feedback page.

Not a high priority....

During today's meeting it was brought up that the geographic features visible through the colored county/tract geometry makes it a little hard to see the color of the geoemtry. We could either add other tilemaps with less or different features, or simply make the geometry opaque as it is on the KY Cancer InFocus site.

Having a reset button for the entire map would be helpful. Thanks!

Per our discussion today, it would be good to be able to filter the data based on two metrics at once, for example % college graduates vs. % cancer incidence.

We don't currently have the data to do an actual intersection of these measures at the moment. Getting access to this data is part 1 of this issue.

Part 2 is a less complex (and perhaps less useful) solution that we can do with the data we have right now. We can essentially display two metrics side by side. The map counties/tracts will always have to be colored only by a single measure, but we can integrate other measures in other ways. The user would select multiple measures and provide criteria for them, e.g. "show me counties with > 20% college grads and < 5% breast cancer rates". Then our UI could either add markers for each criterion at the center of each county/tract, or, the county/tract could be binary on/off (visible/invisible) based on whether it meets all of the criteria. The former will probably more visually difficult to read, especially with more dense areas of counties/tracts, but could show each criteria as a separate marker, so you'd know exactly which criteria were met. With the latter, you only know if all the criteria were met or not.

Both parts of this issue are considered an enhancement, and we're holding off on starting to implement them for now.

Split from #15

Nice to have absolute counts along with % rates (fix: AAC column in dataset?)

• Say where sources came from. Defer to CancerInFocus's sources page rather than duplicating data.
• Specify timespans over which rates are calculated.
• Nice to have absolute counts along with % rates (fix: AAC column in dataset?)
• Add cancer centers, etc. into overlays with comprehensive data (address, hours, etc.)
• Radon data?
• Investigate bug where downloading data produces empty sheet.
• Add support instructions for non-technical audience

One user had an issue connecting to the site. Windows 11, Chrome/Edge/Firefox/etc. Page was not connecting at all, i.e.:

Pinging both coe-ecco.org and api.coe-ecco.org from the command line was successful, but when trying curl, we received this:

C:\Users\dewright>curl -v coe-ecco.org
*   Trying 75.2.60.5:80...
* Connected to coe-ecco.org (75.2.60.5) port 80
> GET / HTTP/1.1
> Host: coe-ecco.org
> User-Agent: curl/8.4.0
> Accept: */*
>
< HTTP/1.1 503 Service Unavailable
< Content-Type: text/html; charset=UTF-8
< Content-Length: 10522
< Connection: close
< P3P: CP="CAO PSA OUR"
< Expires: Thu, 01 Jan 1970 00:00:00 GMT
< Cache-Control: no-store, no-cache, must-revalidate, post-check=0, pre-check=0
< Pragma: no-cache
<
<html>
<head>
<title>Web Page Security Block</title>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<META HTTP-EQUIV="PRAGMA" CONTENT="NO-CACHE">
<meta name="viewport" content="initial-scale=1.0">
<style>
  omitted
</style>
</head>
<body bgcolor="#fff">
<div id="content">
  <center><img src="omitted"/>
<h1>Web Page Blocked CDPHE</h1>
</center>
<p><b>Access to the web page you were trying to visit has been blocked in accordance with OIT policy.</b></p>
<p><b>Due to security concerns, this page has been blocked from the State network.  If you believe this website is needed for business purposes, please click <a href="https://servicehub.colorado.gov/" target="_blank">here</a> and follow the instructions below to request access to the website.   You will be directed to the ServiceHub Customer Support Portal.</b></p>
<p><b>Instructions:</b></p>
<ol>
<b>
<li>If needed, for guidance on ServiceHub login credentials click <q>What are my username and password?</q></li>
<li>Once logged in, click <q>Browse Services </q> and search for <q>Unblock Website</q>. Click the "Request" button under the resulting search and complete the ticket template. </li>
<li>Please be sure to include  the website address, time of block, a brief business use justification and attach a screenshot of this block page.</li>
<li>When the Unblock Website template is completed, click on Submit.</li>
<li>For additional assistance Google Chat: OIT_ServiceDesk_HQ or Call: 303.239.HELP (4357)</li>
</b>
</ol>
<p><b>User:</b> cdphe.coloradostate.gov\dewright </p>
<p><b>URL:</b> coe-ecco.org/ </p>
<p><b>Category:</b> newly-registered-domain </p>
<p>OUTBOUND-WEB </p>
</div>
</body>
</html>
* Closing connection

Does it make sense to get a deployment up? If so, how do you all want to deploy?

In tables that have stratification variables, aka "factors", like age or sex, the table sometimes includes rows labeled "All" for the factor, and sometimes it doesn't. This poses a problem when you don't want to stratify by a factor, but "All" isn't present.

I could imagine a way to compute "All", but it would differ for different types of values:

• for counts, "All" would be the sum of the counts
• for percentages, "All" would be the average across the percentages?
• for ordinal values (e.g., the trends column in the SCP dataset, where the values can be assigned numbers by their ordering), it could also the average or perhaps the mode (the most common value)
• we don't currently handle showing categorical values on the map (i.e, discrete, non-ordered values like "race"), and I don't know how you'd summarize them...perhaps it'd be the mode, too?

Presumably if we computed "All" ourselves we'd have to exclude the rows with "All" in them, otherwise they'd throw off the results.

Hi -

I'm noticing that when I populate the incidence and mortality, the default age is "<50". Is it possible to change this to "All Ages"? Don't know if this is only on my end but hoping ya'll could double-check. Thanks!

Name

Cancer Disparities Index

URL

https://cohealthviz.dphe.state.co.us/t/HSEBPublic/views/CCPDDiseaseDisparityCountyIndexRanking/CCPDRankings?%3Aembed=y&%3Adisplay_count=no

Description

As I understand it, the Colorado Department of Public Health and Environment (CDPHE) used a PCA-based approach to come up with health disparities indices. A high-level description is available here: https://cohealthviz.dphe.state.co.us/t/HSEBPublic/views/CCPDDiseaseDisparityCountyIndexRanking/CCPDRankings?%3Aembed=y&%3Adisplay_count=no

Relevance

The idea is to provide a map of that highlights disparities at the county level.

Anything else?

This is an Excel file prepared by Amy Mellies when she was working in PHSR with COE that includes similar information to that available from the CDPHE website.

DT-COE Cancer Disparity Indices and Treatment Locations by County.xlsx

Maybe this can just be a few buttons, like "Suggest A New Data Source (w/ GitHub Account)" and "Contact" (direct email for issues, questions, etc).

We want to be able to filter data by race/ethnicity and by sex. Example image from Cancer InFocus:

Here is a screenshot of the site. It is simple, but it is probably not a bad place to start.

The user in question said he was using iOS/Safari version 16.x (don't remember the minor version. I've used Testing back to iOS version 15 on BrowserStack, I get:

[Error] SyntaxError: Invalid regular expression: invalid group specifier name
	RegExp (index-Bmiwqowh.js:823:1872362)
	npe (index-Bmiwqowh.js:823:1872362)
	map
	Module Code (index-Bmiwqowh.js:823:1872308)
	evaluate
	moduleEvaluation
	(anonymous function)
	promiseReactionJob

Googling, it looks like Safari didn't support regex lookbehinds until very recently:

https://stackoverflow.com/a/51568859/2180570
https://caniuse.com/js-regexp-lookbehind

Working on a fix now.

I'm getting data on incidence and mortality broken down by below. The data will be available as a long-form table (should be a few million rows). I'll share when I'm done. The data look like:

Some quotes from Teams chat:

[6/13 3:55 PM] Lowery, Jan
Hi All, a few updates and requests: we shared ECCO with our Community Advisory Council last month and asked for feedback. So far, only 10 people have filled out our feedback survey. hoping we get more. .......... We also heard from a few folks that they want to see Colorado data - so cancer incidence/mortality for Colorado as a whole - I believe the Profiles data has this, question is how to display on map? Maybe within legend? or maybe separate box? This applies to cancer rates and also risk factors, e.g. smoking prevalence, % of people up to date with screening. Any ideas on this?

[6/13 7:39 PM] Rubinetti, Vincent
regarding the state-wide info. if i'm understanding correctly, this would basically just be a single number for the currently selected measure? e.g. sociodemographics -> total population would show a single number for the entire population of colorado? if so, yeah we'd probably just put that total right in one of the legends. if you go into "customization" and check "extra info", another legend pops up with min/max/avg/etc. we could add "total" or "overall" to that. i'd also have "extra info" checked by default then.
like 2
[6/13 7:43 PM] Rubinetti, Vincent
for some types of measures that's probably as simple as totaling all the county/tract values. for others like %s or indices, it's probably more complicated or not relevant. if it's something we need to capture from the ingested data rather than calculate ourselves, we'll have to wait until Faisal is back next week

Summary: make "extra info" checkbox (under "Customization") checked by default, and add "Total" or "Overall" value to the legend along with the min/max/mean/median that's already there. The legend is basically showing overall values (for the whole state of Colorado) for the selected measure.

I could do a total very easily from the frontend, right where I'm already doing the average and median calculations. I could use the "unit type" Faisal added to each measure to only show the total when it makes sense (e.g. "total population" or "cancer mortality -> all cancer sites") and hide it when it doesn't (e.g. "annual labor force participation %").

From Jake Walsh's email:

The second idea is adding the state legislative maps to ECCO. The maps would be put under the Locations menu and a user could select either the state House of Representatives or state Senate map. The selected map would then be overlayed over either the county or census tract maps. I already have the shapefiles for both district maps and an Excel sheet that has all of the legislators' names and contact information. I just want to see if it's possible to import these maps and overlay them in ECCO.

Since as Jake mentioned we don't have much data on a district level, there's no reason for it to be a geometry type alongside the county and tract maps. Essentially, the district regions are the data and should be superimposed over a selected measure for a county or tract.

Data required for implementation:

• shapefiles of the district maps
• spreadsheet with additional information for each district, e.g. the representative's name and contact info (displayed in a popup for the region)

https://gist.github.com/seandavi/3d3136aaf66b0b58e2f1906085415312

1. After selecting a county, the pop-up window pulls up too much information, and is generally confusing. What is the difference between the Value and the Age-Adj Count? Which number represents the rate? As I look through, it seems that the Value reflects the true rate, so it may be as easy as getting rid of the Age-adj Count label... Example below.

2. We do not need to include the FIPS number anywhere; this can be removed from the pop-up.

Currently, measures and measure categories like the Cancer Disparity Index are a bit opaque: it's unclear what the measure means. It would be good to have a simple description of measures that might be confusing, and more text (perhaps on a separate "glossary" page) that more fully describes the measure.

We'll likely display the short explanation in the "no data" explanation box, possibly with a link to the longer description on the "glossary" page.

Todo:

• disparity measures writeup, social vs. cancer incidence/mortality
• add other measures that we want to describe in detail

Currently, groupings of data in ECCO can start at 0. However, for privacy concerns, if there are values that are less than 5 within cancer incidence/mortality, we need to suppress this data. Example from Cancer InFocus below.

It would also be helpful to have as specific as possible the legends and groupings, instead of numbers that are rounded down/up. The picture below is an example.

@vincerubinetti @falquaddoomi @janolowery Let me know if this is the format you want to get feedback:

• The map shows screening sites for "lung cancer screening" across cancers (e.g., female breast cancer). Is it possible to add the "other cancer screening sites" (e.g., Women Wellness Connection for Breast and Cervical Cancer) in our Website. https://medschool.cuanschutz.edu/colorado-cancer-center/community/CommunityOutreachEngagement/Colorado-Cancer-map
• Do you need a data source to link?

From Jan (paraphrased): It can be disorienting to see the colors change as you select different measure categories, measures, and geographic groupings (county vs. tract). Since the colors depend on the dataset, you can't use specific colors as a means to compare across different measures.

Background: The map colors are generated by taking the minimum and maximum values (aka the "extent") of the currently-visible dataset, then generating a set number of steps between. Some measures are percentages, while others are literal values.

Options:

• scale each measure separately across its county and tract values
  • issues: doesn't provide color consistency across measures
• find a global min, max across all measures
  • issues: wouldn't work well for percentages vs literal values
• allow the user to specify that they want an absolute scale, and then to specify the min and max values for that scale
  • issues: user might become disoriented switching between, e.g. percentages vs. literals, or measures with wildly different extents

Some things that would help flesh out this issue and make it more concrete:

• examples of pairs of measures where the difference is extreme
• user stories, i.e. specific users who would want this feature
• other options on how to present the data

After pulling the new Cancer InFocus data in late February, I'd noticed that they'd introduced new "disparities" tables at the county and tract levels. The tables contain the following metrics, where the english description that follows each field was gathered from the dropdown on https://cancerinfocus.uky.edu/:

• Living Below Poverty (Black): "Black Population Living Below Poverty"
• Uninsured (Black): "Black Population without Health Insurance"
• Uninsured Children: "Children without Health Insurance"
• Economic Segregation: "Economic Segregation"
• Gender Pay Gap: "Gender Pay Gap"
• Living Below Poverty (Hispanic): "Hispanic Population Living Below Poverty"
• Uninsured (Hispanic): "Hispanic Population without Health Insurance"
• Gini Coefficient: "Income Inequality (Gini Coefficient)"
• Racial Economic Segregation: "Racial Economic Segregation"
• Racial Segregation: "Racial Segregation"
• Lack_English_Prof: (missing on https://cancerinfocus.uky.edu/)

Lack_English_Prof is an interesting case, since it already appears under the "Sociodemographics" category. I'm unsure if we should include it under "Disparities", too, or if we should just rely on the "Sociodemographics" version, especially considering it's not on the KY site.

It was relatively easy to integrate these new tables into our system; I'll be submitting that work in a PR shortly.

Note that this data is different from the data described in #27; this data appears to be metrics regarding socioeconomic disparities in general, not cancer-specific disparities.

Not sure if this is only for lung & bronchus, but when we hit download next to the measure, it comes up with a blank Excel. When we hit the download button next to the measure category, the data opens up. Hopefully just an easy bug to fix. Thanks!

Split from #15

I believe the CancerInFocus data is from the last 2-5 years or something like that? Might be best to defer external pages for that info?

Split from #15

The data for CancerInFocus are available for the entire US as a Zip from here:

I suspect that the download URL may change over time, so scraping to get the URL may be required. You'll get multiple "grains" of data (state, county, etc.). There is also a "long" and "wide" version for each resource.