Proxy Detection Logic Flaw?

Following up on my previous post about my public IP address tool, I've been trying to refine the proxy detection logic, and I'm hitting a wall. Specifically, I'm getting inconsistent results when trying to differentiate between legitimate VPN users and actual open proxies or residential proxies being abused.

My current approach involves checking common HTTP headers (X-Forwarded-For, Via, Proxy-Connection) and comparing the observed IP's geolocation with the client's reported locale. However, this often flags legitimate VPNs as proxies or misses sophisticated residential proxies entirely.

Here's a simplified version of the check that's causing issues:

function detectProxy(headers, clientIp) {
  if (headers['X-Forwarded-For'] || headers['Via'] || headers['Proxy-Connection']) {
    return true; // Simple header check
  }
  // More complex IP reputation/geolocation checks follow...
  return false;
}

// Example problematic output:
// Request from IP: 192.0.2.1 (VPN user in USA)
// Headers: {'User-Agent': '...', 'X-Forwarded-For': '198.51.100.1'}
// detectProxy(headers, '192.0.2.1') -> true (FALSE POSITIVE)

// Request from IP: 203.0.113.1 (Residential proxy in UK)
// Headers: {'User-Agent': '...'}
// detectProxy(headers, '203.0.113.1') -> false (FALSE NEGATIVE)

I'm looking for more robust methods or algorithms for accurate proxy detection. How do you guys handle this without generating too many false positives for privacy-conscious users?

Help a brother out please...